frame.ts

/**
*  @license
* Copyright 2022 JsData. All rights reserved.
*
* This source code is licensed under the MIT license found in the
* LICENSE file in the root directory of this source tree.

* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
* ==========================================================================
*/
import dummyEncode from "../transformers/encoders/dummy.encoder";
import { variance, std, median, mode, mean } from 'mathjs';
import tensorflow from '../shared/tensorflowlib'
import { DATA_TYPES } from '../shared/defaults';
import { _genericMathOp } from "./math.ops";
import Groupby from '../aggregators/groupby';
import ErrorThrower from "../shared/errors"
import { _iloc, _loc } from "./indexing";
import Utils from "../shared/utils"
import NDframe from "./generic";
import { table } from "table";
import Series from './series';
import {
    ArrayType1D,
    ArrayType2D,
    DataFrameInterface,
    BaseDataOptionType,
    IPlotlyLib,
} from "../shared/types";
import { PlotlyLib } from "../../danfojs-base/plotting";

const utils = new Utils();

/**
 * Two-dimensional ndarray with axis labels.
 * The object supports both integer- and label-based indexing and provides a host of methods for performing operations involving the index.
 * Operations between DataFrame (+, -, /, , *) align values based on their associated index values– they need not be the same length.
 * @param data 2D Array, JSON, Tensor, Block of data.
 * @param options.index Array of numeric or string names for subseting array. If not specified, indexes are auto generated.
 * @param options.columns Array of column names. If not specified, column names are auto generated.
 * @param options.dtypes Array of data types for each the column. If not specified, dtypes are/is inferred.
 * @param options.config General configuration object for extending or setting NDframe behavior.      
 */
export default class DataFrame extends NDframe implements DataFrameInterface {
    [key: string]: any
    constructor(data: any, options: BaseDataOptionType = {}) {
        const { index, columns, dtypes, config } = options;
        super({ data, index, columns, dtypes, config, isSeries: false });
        this.$setInternalColumnDataProperty();
    }

    /**
     * Maps all column names to their corresponding data, and return them as Series objects.
     * This makes column subsetting works. E.g this can work ==> `df["col1"]`
     * @param column Optional, a single column name to map
     */
    private $setInternalColumnDataProperty(column?: string) {
        const self = this;
        if (column && typeof column === "string") {
            Object.defineProperty(self, column, {
                get() {
                    return self.$getColumnData(column)
                },
                set(arr: ArrayType1D | Series) {
                    self.$setColumnData(column, arr);
                }
            })
        } else {
            const columns = this.columns;
            for (let i = 0; i < columns.length; i++) {
                const column = columns[i];
                Object.defineProperty(this, column, {
                    get() {
                        return self.$getColumnData(column)
                    },
                    set(arr: ArrayType1D | Series) {
                        self.$setColumnData(column, arr);
                    }
                })
            }
        }

    }

    /**
     * Returns the column data from the DataFrame by column name. 
     * @param column column name to get the column data
     * @param returnSeries Whether to return the data in series format or not. Defaults to true
     */
    private $getColumnData(column: string, returnSeries: boolean = true) {
        const columnIndex = this.columns.indexOf(column)

        if (columnIndex == -1) {
            ErrorThrower.throwColumnNotFoundError(this)
        }

        const dtypes = [this.$dtypes[columnIndex]]
        const index = [...this.$index]
        const columns = [column]
        const config = { ...this.$config }

        if (this.$config.isLowMemoryMode) {
            const data: ArrayType1D = []
            for (let i = 0; i < this.values.length; i++) {
                const row: any = this.values[i];
                data.push(row[columnIndex])
            }
            if (returnSeries) {
                return new Series(data, {
                    dtypes,
                    index,
                    columns,
                    config
                })
            } else {
                return data
            }

        } else {
            const data = this.$dataIncolumnFormat[columnIndex]
            if (returnSeries) {
                return new Series(data, {
                    dtypes,
                    index,
                    columns,
                    config
                })
            } else {
                return data
            }
        }

    }


    /**
     * Updates the internal column data via column name.
     * @param column The name of the column to update.
     * @param arr The new column data
     */
    private $setColumnData(column: string, arr: ArrayType1D | Series): void {

        const columnIndex = this.$columns.indexOf(column)

        if (columnIndex == -1) {
            throw new Error(`ParamError: column ${column} not found in ${this.$columns}. If you need to add a new column, use the df.addColumn method. `)
        }

        let colunmValuesToAdd: ArrayType1D

        if (arr instanceof Series) {
            colunmValuesToAdd = arr.values as ArrayType1D
        } else if (Array.isArray(arr)) {
            colunmValuesToAdd = arr;
        } else {
            throw new Error("ParamError: specified value not supported. It must either be an Array or a Series of the same length")
        }

        if (colunmValuesToAdd.length !== this.shape[0]) {
            ErrorThrower.throwColumnLengthError(this, colunmValuesToAdd.length)
        }

        if (this.$config.isLowMemoryMode) {
            //Update row ($data) array
            for (let i = 0; i < this.$data.length; i++) {
                (this.$data as any)[i][columnIndex] = colunmValuesToAdd[i]
            }
            //Update the dtypes
            this.$dtypes[columnIndex] = utils.inferDtype(colunmValuesToAdd)[0]
        } else {
            //Update row ($data) array
            for (let i = 0; i < this.values.length; i++) {
                (this.$data as any)[i][columnIndex] = colunmValuesToAdd[i]
            }
            //Update column ($dataIncolumnFormat) array since it's available in object
            (this.$dataIncolumnFormat as any)[columnIndex] = arr

            //Update the dtypes
            this.$dtypes[columnIndex] = utils.inferDtype(colunmValuesToAdd)[0]
        }

    }

    /**
     * Return data with missing values removed from a specified axis
     * @param axis 0 or 1. If 0, column-wise, if 1, row-wise
    */
    private $getDataByAxisWithMissingValuesRemoved(axis: number): Array<number[]> {
        const oldValues = this.$getDataArraysByAxis(axis);
        const cleanValues = [];
        for (let i = 0; i < oldValues.length; i++) {
            const values = oldValues[i] as number[]
            cleanValues.push(utils.removeMissingValuesFromArray(values) as number[]);
        }
        return cleanValues;
    }

    /**
     * Return data aligned to the specified axis. Transposes the array if needed.
     * @param axis 0 or 1. If 0, column-wise, if 1, row-wise
    */
    private $getDataArraysByAxis(axis: number): ArrayType2D {
        if (axis === 1) {
            return this.values as ArrayType2D
        } else {
            let dfValues;
            if (this.config.isLowMemoryMode) {
                dfValues = utils.transposeArray(this.values) as ArrayType2D
            } else {
                dfValues = this.$dataIncolumnFormat as ArrayType2D
            }
            return dfValues
        }
    }

    /* 
    * checks if DataFrame is compactible for arithmetic operation
    * compatible Dataframe must have only numerical dtypes
    **/
    private $frameIsNotCompactibleForArithmeticOperation() {
        const dtypes = this.dtypes;
        const str = (element: any) => element == "string";
        return dtypes.some(str)
    }

    /**
     * Return Tensors in the right axis for math operations.
     * @param other DataFrame or Series or number or array
     * @param axis 0 or 1. If 0, column-wise, if 1, row-wise
     * */
    private $getTensorsForArithmeticOperationByAxis(
        other: DataFrame | Series | number | Array<number>,
        axis: number
    ) {
        if (typeof other === "number") {
            return [this.tensor, tensorflow.scalar(other)];
        } else if (other instanceof DataFrame) {
            return [this.tensor, other.tensor];
        } else if (other instanceof Series) {
            if (axis === 0) {
                return [this.tensor, tensorflow.tensor2d(other.values as Array<number>, [other.shape[0], 1])];
            } else {
                return [this.tensor, tensorflow.tensor2d(other.values as Array<number>, [other.shape[0], 1]).transpose()];
            }
        } else if (Array.isArray(other)) {
            if (axis === 0) {
                return [this.tensor, tensorflow.tensor2d(other, [other.length, 1])];
            } else {
                return [this.tensor, tensorflow.tensor2d(other, [other.length, 1]).transpose()];
            }
        } else {
            throw new Error("ParamError: Invalid type for other parameter. other must be one of Series, DataFrame or number.")
        }

    }

    /**
     * Returns the dtype for a given column name
     * @param column 
     */
    private $getColumnDtype(column: string): string {
        const columnIndex = this.columns.indexOf(column)
        if (columnIndex === -1) {
            throw Error(`ColumnNameError: Column "${column}" does not exist`)
        }
        return this.dtypes[columnIndex]
    }

    private $logicalOps(tensors: typeof tensorflow.Tensor[], operation: string) {
        let newValues: number[] = []

        switch (operation) {
            case 'gt':
                newValues = tensors[0].greater(tensors[1]).arraySync() as number[]
                break;
            case 'lt':
                newValues = tensors[0].less(tensors[1]).arraySync() as number[]
                break;
            case 'ge':
                newValues = tensors[0].greaterEqual(tensors[1]).arraySync() as number[]
                break;
            case 'le':
                newValues = tensors[0].lessEqual(tensors[1]).arraySync() as number[]
                break;
            case 'eq':
                newValues = tensors[0].equal(tensors[1]).arraySync() as number[]
                break;
            case 'ne':
                newValues = tensors[0].notEqual(tensors[1]).arraySync() as number[]
                break;

        }

        const newData = utils.mapIntegersToBooleans(newValues, 2)
        return new DataFrame(
            newData,
            {
                index: [...this.index],
                columns: [...this.columns],
                dtypes: [...this.dtypes],
                config: { ...this.config }
            })
    }

    private $MathOps(tensors: typeof tensorflow.Tensor[], operation: string, inplace: boolean) {
        let tensorResult

        switch (operation) {
            case 'add':
                tensorResult = tensors[0].add(tensors[1])
                break;
            case 'sub':
                tensorResult = tensors[0].sub(tensors[1])
                break;
            case 'pow':
                tensorResult = tensors[0].pow(tensors[1])
                break;
            case 'div':
                tensorResult = tensors[0].div(tensors[1])
                break;
            case 'divNoNan':
                tensorResult = tensors[0].divNoNan(tensors[1])
                break;
            case 'mul':
                tensorResult = tensors[0].mul(tensors[1])
                break;
            case 'mod':
                tensorResult = tensors[0].mod(tensors[1])
                break;
        }

        if (inplace) {
            const newData = tensorResult?.arraySync() as ArrayType2D
            this.$setValues(newData)
        } else {
            return new DataFrame(
                tensorResult,
                {
                    index: [...this.index],
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                })

        }
    }

    /**
    * Purely integer-location based indexing for selection by position.
    * ``.iloc`` is primarily integer position based (from ``0`` to
    * ``length-1`` of the axis), but may also be used with a boolean array.
    * 
    * @param rows Array of row indexes
    * @param columns Array of column indexes
    *  
    * Allowed inputs are in rows and columns params are:
    * 
    * - An array of single integer, e.g. ``[5]``.
    * - A list or array of integers, e.g. ``[4, 3, 0]``.
    * - A slice array string with ints, e.g. ``["1:7"]``.
    * - A boolean array.
    * - A ``callable`` function with one argument (the calling Series or
    * DataFrame) and that returns valid output for indexing (one of the above).
    * This is useful in method chains, when you don't have a reference to the
    * calling object, but would like to base your selection on some value.
    * 
    * ``.iloc`` will raise ``IndexError`` if a requested indexer is
    * out-of-bounds.
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
    * const df2 = df.iloc({ rows: [1], columns: ["A"] })
    * ```
    */
    iloc({ rows, columns }: {
        rows?: Array<string | number | boolean> | Series,
        columns?: Array<string | number>
    }): DataFrame {
        return _iloc({ ndFrame: this, rows, columns }) as DataFrame;
    }


    /**
     * Access a group of rows and columns by label(s) or a boolean array.
     * ``loc`` is primarily label based, but may also be used with a boolean array.
     * 
     * @param rows Array of row indexes
     * @param columns Array of column indexes
     * 
     * Allowed inputs are:
     * 
     * - A single label, e.g. ``["5"]`` or ``['a']``, (note that ``5`` is interpreted as a 
     *   *label* of the index, and **never** as an integer position along the index).
     * 
     * - A list or array of labels, e.g. ``['a', 'b', 'c']``.
     * 
     * - A slice object with labels, e.g. ``["a:f"]``. Note that start and the stop are included
     * 
     * - A boolean array of the same length as the axis being sliced,
     * e.g. ``[True, False, True]``.
     * 
     * - A ``callable`` function with one argument (the calling Series or
     * DataFrame) and that returns valid output for indexing (one of the above)
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
    * const df2 = df.loc({ rows: [1], columns: ["A"] })
    * ```
    */
    loc({ rows, columns }: {
        rows?: Array<string | number | boolean> | Series,
        columns?: Array<string>
    }): DataFrame {
        return _loc({ ndFrame: this, rows, columns }) as DataFrame
    }

    /**
     * Prints DataFrame to console as a formatted grid of row and columns.
    */
    toString(): string {
        const maxRow = this.config.getMaxRow;
        const maxColToDisplayInConsole = this.config.getTableMaxColInConsole;

        // let data;
        const dataArr: ArrayType2D = [];
        const colLen = this.columns.length;

        let header = [];

        if (colLen > maxColToDisplayInConsole) {
            //truncate displayed columns to fit in the console
            let firstFourcolNames = this.columns.slice(0, 4);
            let lastThreecolNames = this.columns.slice(colLen - 3);
            //join columns with truncate ellipse in the middle
            header = ["", ...firstFourcolNames, "...", ...lastThreecolNames];

            let subIdx: Array<number | string>
            let firstHalfValues: ArrayType2D
            let lastHalfValueS: ArrayType2D

            if (this.values.length > maxRow) {
                //slice Object to show [max_rows]
                let dfSubset1 = this.iloc({
                    rows: [`0:${maxRow}`],
                    columns: ["0:4"]
                });

                let dfSubset2 = this.iloc({
                    rows: [`0:${maxRow}`],
                    columns: [`${colLen - 3}:`]
                });

                subIdx = this.index.slice(0, maxRow);
                firstHalfValues = dfSubset1.values as ArrayType2D
                lastHalfValueS = dfSubset2.values as ArrayType2D

            } else {
                let dfSubset1 = this.iloc({ columns: ["0:4"] });
                let dfSubset2 = this.iloc({ columns: [`${colLen - 3}:`] });

                subIdx = this.index.slice(0, maxRow);
                firstHalfValues = dfSubset1.values as ArrayType2D
                lastHalfValueS = dfSubset2.values as ArrayType2D
            }

            // merge subset 
            for (let i = 0; i < subIdx.length; i++) {
                const idx = subIdx[i];
                const row = [idx, ...firstHalfValues[i], "...", ...lastHalfValueS[i]]
                dataArr.push(row);
            }

        } else {
            //display all columns
            header = ["", ...this.columns]
            let subIdx
            let values: ArrayType2D
            if (this.values.length > maxRow) {
                //slice Object to show a max of [max_rows]
                let data = this.iloc({ rows: [`0:${maxRow}`] });
                subIdx = data.index;
                values = data.values as ArrayType2D
            } else {
                values = this.values as ArrayType2D
                subIdx = this.index;
            }

            // merge subset 
            for (let i = 0; i < subIdx.length; i++) {
                const idx = subIdx[i];
                const row = [idx, ...values[i]];
                dataArr.push(row);
            }
        }


        const columnsConfig: any = {};
        columnsConfig[0] = { width: 10 }; //set column width for index column

        for (let index = 1; index < header.length; index++) {
            columnsConfig[index] = { width: 17, truncate: 16 };
        }

        const tableData: any = [header, ...dataArr]; //Adds the column names to values before printing

        return table(tableData, { columns: columnsConfig, ...this.config.getTableDisplayConfig });
    }

    /**
      * Returns the first n values in a DataFrame
      * @param rows The number of rows to return
      * @example
      * ```
      * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
      * const df2 = df.head(1)
      * ```
    */
    head(rows: number = 5): DataFrame {

        if (rows <= 0) {
            throw new Error("ParamError: Number of rows cannot be less than 1")
        }
        if (this.shape[0] <= rows) {
            return this.copy()
        }
        if (this.shape[0] - rows < 0) {
            throw new Error("ParamError: Number of rows cannot be greater than available rows in data")
        }

        return this.iloc({ rows: [`0:${rows}`] })
    }

    /**
      * Returns the last n values in a DataFrame
      * @param rows The number of rows to return
      * @example
      * ```
      * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
      * const df2 = df.tail(1)
      * ```
    */
    tail(rows: number = 5): any {

        if (rows <= 0) {
            throw new Error("ParamError: Number of rows cannot be less than 1")
        }
        if (this.shape[0] <= rows) {
            return this.copy()
        }
        if (this.shape[0] - rows < 0) {
            throw new Error("ParamError: Number of rows cannot be greater than available rows in data")
        }

        rows = this.shape[0] - rows
        return this.iloc({ rows: [`${rows}:`] })
    }

    /**
     * Gets n number of random rows in a dataframe. Sample is reproducible if seed is provided.
     * @param num The number of rows to return. Default to 5.
     * @param options.seed An integer specifying the random seed that will be used to create the distribution.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = await df.sample(1)
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = await df.sample(1, { seed: 1 })
     * ```
    */
    async sample(num = 5, options?: { seed?: number }): Promise<DataFrame> {
        const { seed } = { seed: 1, ...options }

        if (num > this.shape[0]) {
            throw new Error("ParamError: Sample size cannot be bigger than number of rows");
        }
        if (num <= 0) {
            throw new Error("ParamError: Sample size cannot be less than 1");
        }

        const shuffledIndex = await tensorflow.data.array(this.index).shuffle(num, `${seed}`).take(num).toArray();
        const df = this.iloc({ rows: shuffledIndex });
        return df;
    }

    /**
     * Return Addition of DataFrame and other, element-wise (binary operator add).
     * @param other DataFrame, Series, Array or Scalar number to add
     * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.add(1)
     * df2.print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * df.add([1, 2], { axis: 0, inplace: true})
     * df.print()
     * ```
    */
    add(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    add(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: add operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "add", inplace)
    }

    /**
     * Return substraction between DataFrame and other.
     * @param other DataFrame, Series, Array or Scalar number to substract from DataFrame
     * @param options.axis 0 or 1. If 0, compute the subtraction column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.sub(1)
     * df2.print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * df.sub([1, 2], { axis: 0, inplace: true})
     * df.print()
     *  ```
    */
    sub(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    sub(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: sub operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "sub", inplace)


    }
    /**
     * Return multiplciation between DataFrame and other.
     * @param other DataFrame, Series, Array or Scalar number to multiply with.
     * @param options.axis 0 or 1. If 0, compute the multiplication column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.mul(2)
     * df2.print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * df.mul([1, 2], { axis: 0, inplace: true})
     * df.print()
     * ```
    */
    mul(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    mul(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: mul operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }
        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "mul", inplace)


    }

    /**
     * Return division of DataFrame with other.
     * @param other DataFrame, Series, Array or Scalar number to divide with.
     * @param options.axis 0 or 1. If 0, compute the division column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.div(2)
     * df2.print()
     * ```
    */
    div(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    div(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: div operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "div", inplace)


    }

    /**
     * Return division of DataFrame with other, returns 0 if denominator is 0.
     * @param other DataFrame, Series, Array or Scalar number to divide with.
     * @param options.axis 0 or 1. If 0, compute the division column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.divNoNan(2)
     * df2.print()
     * ```
    */
    divNoNan(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    divNoNan(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: div operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "divNoNan", inplace)


    }

    /**
     * Return DataFrame raised to the power of other.
     * @param other DataFrame, Series, Array or Scalar number to to raise to.
     * @param options.axis 0 or 1. If 0, compute the power column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.pow(2)
     * df2.print()
     * ```
    */
    pow(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    pow(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: pow operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "pow", inplace)


    }

    /**
     * Return modulus between DataFrame and other.
     * @param other DataFrame, Series, Array or Scalar number to modulus with.
     * @param options.axis 0 or 1. If 0, compute the mod column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * const df2 = df.mod(2)
     * df2.print()
     * ```
    */
    mod(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    mod(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: mod operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$MathOps(tensors, "mod", inplace)

    }

    /**
     * Return mean of DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the mean column-wise, if 1, row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * df.mean().print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B'] })
     * df.mean({ axis: 0 }).print()
     * ```
    */
    mean(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: mean operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const resultArr = newData.map(arr => arr.reduce((a, b) => a + b, 0) / arr.length)
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }
    }

    /**
     * Return median of DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the median column-wise, if 1, row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2, 4], [3, 4, 5], [6, 7, 8]], { columns: ['A', 'B', 'C'] });
     * df.median().print()
     * ```
    */
    median(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: median operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const resultArr = newData.map(arr => median(arr))
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }

    }

    /**
     * Return mode of DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the mode column-wise, if 1, row-wise. Defaults to 1
     * @param options.keep If there are more than one modes, returns the mode at position [keep]. Defaults to 0
     * @example
     * ```
     * const df = new DataFrame([[1, 2, 4], [3, 4, 5], [6, 7, 8]], { columns: ['A', 'B', 'C'] });
     * df.mode().print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2, 4], [3, 4, 5], [6, 7, 8]], { columns: ['A', 'B', 'C'] });
     * df.mode({ keep: 1 }).print()
     * ```
    */
    mode(options?: { axis?: 0 | 1, keep?: number }): Series {
        const { axis, keep } = { axis: 1, keep: 0, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: mode operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const resultArr = newData.map(arr => {
            const tempMode = mode(arr)
            if (tempMode.length === 1) {
                return tempMode[0]
            } else {
                return tempMode[keep]
            }
        })
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }

    }

    /**
     * Return minimum of values in a DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the minimum value column-wise, if 1, row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.min().print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.min({ axis: 0 }).print()
     * ```
    */
    min(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: min operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)

        const resultArr = newData.map(arr => {
            let smallestValue = arr[0]
            for (let i = 0; i < arr.length; i++) {
                smallestValue = smallestValue < arr[i] ? smallestValue : arr[i]
            }
            return smallestValue
        })
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }
    }

    /**
     * Return maximum of values in a DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the maximum column-wise, if 1, row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.max().print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.max({ axis: 0 }).print()
     * ```
    */
    max(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: max operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)

        const resultArr = newData.map(arr => {
            let biggestValue = arr[0]
            for (let i = 0; i < arr.length; i++) {
                biggestValue = biggestValue > arr[i] ? biggestValue : arr[i]
            }
            return biggestValue
        })
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }

    }

    /**
     * Return standard deviation of values in a DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the standard deviation column-wise, if 1, row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.std().print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.std({ axis: 0 }).print()
     * ```
    */
    std(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: std operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const resultArr = newData.map(arr => std(arr))
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }

    }

    /**
     * Return variance of values in a DataFrame across specified axis.
     * @param options.axis 0 or 1. If 0, compute the variance column-wise, if 1, add row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.var().print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.var({ axis: 0 }).print()
     * ```
    */
    var(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: var operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const resultArr = newData.map(arr => variance(arr))
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }

    }

    /**
     * Get Less than of dataframe and other, element-wise (binary operator lt).
     * @param other DataFrame, Series, Array or Scalar number to compare with
     * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.lt(2).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.lt([2, 3], { axis: 0 }).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = new Series([2, 3])
     * df.lt(sf, { axis: 1 }).print()
     * ```
    */
    lt(other: DataFrame | Series | number | Array<number>, options?: { axis?: 0 | 1 }): DataFrame {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: lt operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$logicalOps(tensors, "lt")
    }

    /**
     * Returns "greater than" of dataframe and other.
     * @param other DataFrame, Series, Array or Scalar number to compare with
     * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.gt(2).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.gt([2, 3], { axis: 0 }).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = new Series([2, 3])
     * df.gt(sf, { axis: 1 }).print()
     * ```
    */
    gt(other: DataFrame | Series | number | Array<number>, options?: { axis?: 0 | 1 }): DataFrame {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: gt operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$logicalOps(tensors, "gt")
    }

    /**
     * Returns "equals to" of dataframe and other.
     * @param other DataFrame, Series, Array or Scalar number to compare with
     * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.eq(2).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.eq([2, 3], { axis: 0 }).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = new Series([2, 3])
     * df.eq(sf, { axis: 1 }).print()
     * ```
    */
    eq(other: DataFrame | Series | number | Array<number>, options?: { axis?: 0 | 1 }): DataFrame {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: eq operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$logicalOps(tensors, "eq")
    }

    /**
     * Returns "not equal to" of dataframe and other.
     * @param other DataFrame, Series, Array or Scalar number to compare with
     * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.ne(2).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.ne([2, 3], { axis: 0 }).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = new Series([2, 3])
     * df.ne(sf, { axis: 1 }).print()
     * ```
    */
    ne(other: DataFrame | Series | number | Array<number>, options?: { axis?: 0 | 1 }): DataFrame {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: ne operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$logicalOps(tensors, "ne")
    }

    /**
    * Returns "less than or equal to" of dataframe and other.
    * @param other DataFrame, Series, Array or Scalar number to compare with
    * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.le(2).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.le([2, 3], { axis: 0 }).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = new Series([2, 3])
     * df.le(sf, { axis: 1 }).print()
     * ```
    */
    le(other: DataFrame | Series | number | Array<number>, options?: { axis?: 0 | 1 }): DataFrame {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: le operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$logicalOps(tensors, "le")
    }

    /**
    * Returns "greater than or equal to" between dataframe and other.
    * @param other DataFrame, Series, Array or Scalar number to compare with
    * @param options.axis 0 or 1. If 0, add column-wise, if 1, add row-wise
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.ge(2).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.ge([2, 3], { axis: 0 }).print()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = new Series([2, 3])
     * df.ge(sf, { axis: 1 }).print()
     * ```
    */
    ge(other: DataFrame | Series | number | Array<number>, options?: { axis?: 0 | 1 }): DataFrame {
        const { axis } = { axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: ge operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
        return this.$logicalOps(tensors, "ge")
    }

    /**
     * Return number of non-null elements in a Series
     * @param options.axis 0 or 1. If 0, count column-wise, if 1, add row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.count().print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.count({ axis: 0 }).print()
     * ```
    */
    count(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newData = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const resultArr = newData.map(arr => arr.length)
        if (axis === 0) {
            return new Series(resultArr, { index: this.columns });
        } else {
            return new Series(resultArr, { index: this.index });
        }

    }

    /**
     * Return the sum of values across an axis. 
     * @param options.axis 0 or 1. If 0, count column-wise, if 1, add row-wise. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.sum().print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.sum({ axis: 0 }).print()
     * ```
    */
    sum(options?: { axis?: 0 | 1 }): Series {
        const { axis } = { axis: 1, ...options }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const result = this.$getDataByAxisWithMissingValuesRemoved(axis)
        const sumArr = result.map((innerArr) => {
            return innerArr.reduce((a, b) => Number(a) + Number(b), 0)
        })
        if (axis === 0) {
            return new Series(sumArr, {
                index: [...this.columns]
            })
        } else {
            return new Series(sumArr, {
                index: [...this.index]
            })
        }

    }

    /**
     * Return percentage difference of DataFrame with other.
     * @param other DataFrame, Series, Array or Scalar number (positive numbers are preceding rows, negative are following rows) to compare difference with.
     * @param options.axis 0 or 1. If 0, compute the difference column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2, 3, 4, 5, 6], [1, 1, 2, 3, 5, 8], [1, 4, 9, 16, 25, 36]], { columns: ['A', 'B', 'C'] })
     * 
     * // Percentage difference with previous row
     * const df0 = df.pctChange(1)
     * console.log(df0)
     * 
     * // Percentage difference with previous column
     * const df1 = df.pctChange(1, {axis: 0})
     * console.log(df1)
     * 
     * // Percentage difference with previous 3rd previous row
     * const df2 = df.pctChange(3)
     * console.log(df2)
     * 
     * // Percentage difference with following row
     * const df3 = df.pctChange(-1)
     * console.log(df3)
     * 
     * // Percentage difference with another DataFrame
     * const df4 = df.pctChange(df3)
     * console.log(df4)
     * ```
    */
    pctChange(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    pctChange(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: pctChange operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        if (other === 0) {
            return this;
        }

        if (typeof other === "number") {
            let origDF = this.copy() as DataFrame;
            if (axis === 0) {
                origDF = origDF.T;
            }
            const originalTensor = origDF.tensor.clone();
            const unit = new Array(originalTensor.shape[originalTensor.rank - 1]).fill(NaN);
            let pctArray: any[] = originalTensor.arraySync();
            if (other > 0) {
                for (let i = 0; i < other; i++) {
                    pctArray.unshift(unit);
                    pctArray.pop();
                }
            }
            else if (other < 0) {
                for (let i = 0; i > other; i--) {
                    pctArray.push(unit);
                    pctArray.shift();
                }
            }
            const pctTensor = tensorflow.tensor2d(pctArray, originalTensor.shape);
            const pctDF = (this.$MathOps([originalTensor, pctTensor], "divNoNan", inplace) as DataFrame).sub(1);
            if (axis === 0) {
                return pctDF.T;
            }
            return pctDF;
        }

        if (other instanceof DataFrame || other instanceof Series) {
            const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
            const pctDF = (this.$MathOps(tensors, "divNoNan", inplace) as DataFrame).sub(1);
            return pctDF;
        }
    }

    /**
     * Return difference of DataFrame with other.
     * @param other DataFrame, Series, Array or Scalar number (positive numbers are preceding rows, negative are following rows) to compare difference with.
     * @param options.axis 0 or 1. If 0, compute the difference column-wise, if 1, row-wise
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2, 3, 4, 5, 6], [1, 1, 2, 3, 5, 8], [1, 4, 9, 16, 25, 36]], { columns: ['A', 'B', 'C'] })
     * 
     * // Difference with previous row
     * const df0 = df.diff(1)
     * console.log(df0)
     * 
     * // Difference with previous column
     * const df1 = df.diff(1, {axis: 0})
     * console.log(df1)
     * 
     * // Difference with previous 3rd previous row
     * const df2 = df.diff(3)
     * console.log(df2)
     * 
     * // Difference with following row
     * const df3 = df.diff(-1)
     * console.log(df3)
     * 
     * // Difference with another DataFrame
     * const df4 = df.diff(df3)
     * console.log(df4)
     * ```
    */
    diff(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    diff(other: DataFrame | Series | number[] | number, options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { inplace, axis } = { inplace: false, axis: 1, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: diff operation is not supported for string dtypes");
        }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        if (other === 0) {
            return this;
        }

        if (typeof other === "number") {
            let origDF = this.copy() as DataFrame;
            if (axis === 0) {
                origDF = origDF.T;
            }
            const originalTensor = origDF.tensor.clone();
            const unit = new Array(originalTensor.shape[originalTensor.rank - 1]).fill(NaN);
            let diffArray: any[] = originalTensor.arraySync();
            if (other > 0) {
                for (let i = 0; i < other; i++) {
                    diffArray.unshift(unit);
                    diffArray.pop();
                }
            }
            else if (other < 0) {
                for (let i = 0; i > other; i--) {
                    diffArray.push(unit);
                    diffArray.shift();
                }
            }
            const diffTensor = tensorflow.tensor2d(diffArray, originalTensor.shape);
            const diffDF = this.$MathOps([originalTensor, diffTensor], "sub", inplace) as DataFrame;
            if (axis === 0) {
                return diffDF.T;
            }
            return diffDF;
        }

        if (other instanceof DataFrame || other instanceof Series) {
            const tensors = this.$getTensorsForArithmeticOperationByAxis(other, axis);
            return this.$MathOps(tensors, "sub", inplace);
        }
    }

    /**
     * Return the absolute value of elements in a DataFrame. 
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1.0, 2.1], [3.1, 4]], { columns: ['A', 'B']})
     * df.abs().print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1.0, 2], [3.3, 4]], { columns: ['A', 'B']})
     * df.abs({ inplace: true }).print()
     * ```
    */
    abs(options?: { inplace?: boolean }): DataFrame
    abs(options?: { inplace?: boolean }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        const newData = (this.values as number[][]).map(arr => arr.map(val => Math.abs(val)))
        if (inplace) {
            this.$setValues(newData)
        } else {
            return new DataFrame(newData, {
                index: [...this.index],
                columns: [...this.columns],
                dtypes: [...this.dtypes],
                config: { ...this.config }
            })
        }
    }

    /**
     * Rounds all element in the DataFrame to specified number of decimal places.
     * @param dp Number of decimal places to round to. Defaults to 1
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1.12, 2.34], [3.43, 4.0]], { columns: ['A', 'B']})
     * const df2 = df.round(2)
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1.12, 2.34], [3.43, 4.0]], { columns: ['A', 'B']})
     * df.round(2, { inplace: true }).print()
     * ```
    */
    round(dp: number, options?: { inplace: boolean }): DataFrame
    round(dp: number = 1, options?: { inplace: boolean }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        if (this.$frameIsNotCompactibleForArithmeticOperation()) {
            throw Error("TypeError: round operation is not supported for string dtypes");
        }

        if (typeof dp !== "number") {
            throw Error("ParamError: dp must be a number");
        }

        const newData = utils.round(this.values as number[], dp, false);

        if (inplace) {
            this.$setValues(newData)
        } else {
            return new DataFrame(
                newData,
                {
                    index: [...this.index],
                    columns: [...this.columns],
                    config: { ...this.config }
                })
        }
    }

    /**
     * Returns cumulative product accross specified axis.
     * @param options.axis 0 or 1. If 0, count column-wise, if 1, add row-wise. Defaults to 1
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumprod()
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumprod({ axis: 0 })
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.cumprod({ axis: 0, inplace: true }).print()
     * ```
    */
    cumProd(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    cumProd(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { axis, inplace } = { axis: 1, inplace: false, ...options }
        return this.cumOps("prod", axis, inplace);
    }

    /**
     * Returns cumulative sum accross specified axis.
     * @param options.axis 0 or 1. If 0, count column-wise, if 1, add row-wise. Defaults to 1
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumSum()
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumSum({ axis: 0 })
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.cumSum({ axis: 0, inplace: true }).print()
     * ```
    */
    cumSum(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    cumSum(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { axis, inplace } = { axis: 1, inplace: false, ...options }
        return this.cumOps("sum", axis, inplace);
    }

    /**
     * Returns cumulative minimum accross specified axis.
     * @param options.axis 0 or 1. If 0, count column-wise, if 1, add row-wise. Defaults to 1
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumMin()
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumMin({ axis: 0 })
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.cumMin({ axis: 0, inplace: true }).print()
     * ```
    */
    cumMin(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    cumMin(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { axis, inplace } = { axis: 1, inplace: false, ...options }
        return this.cumOps("min", axis, inplace);
    }

    /**
     * Returns cumulative maximum accross specified axis.
     * @param options.axis 0 or 1. If 0, count column-wise, if 1, add row-wise. Defaults to 1
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
          * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumMax()
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.cumMax({ axis: 0 })
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.cumMax({ axis: 0, inplace: true }).print()
     * ```
    */
    cumMax(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame
    cumMax(options?: { axis?: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { axis, inplace } = { axis: 1, inplace: false, ...options }
        return this.cumOps("max", axis, inplace);
    }

    /**
     * Internal helper function for cumulative operation on DataFrame
    */
    private cumOps(ops: string, axis: number, inplace: boolean): DataFrame
    private cumOps(ops: string, axis: number, inplace: boolean): DataFrame | void {
        if (this.dtypes.includes("string")) ErrorThrower.throwStringDtypeOperationError(ops)

        const result = this.$getDataByAxisWithMissingValuesRemoved(axis)

        let newData: ArrayType2D = result.map((sData) => {
            let tempval = sData[0];
            const data = [tempval];

            for (let i = 1; i < sData.length; i++) {
                let currVal = sData[i];
                switch (ops) {
                    case "max":
                        if (currVal > tempval) {
                            data.push(currVal);
                            tempval = currVal;
                        } else {
                            data.push(tempval);
                        }
                        break;
                    case "min":
                        if (currVal < tempval) {
                            data.push(currVal);
                            tempval = currVal;
                        } else {
                            data.push(tempval);
                        }
                        break;
                    case "sum":
                        tempval = (tempval as number) + (currVal as number)
                        data.push(tempval);
                        break;
                    case "prod":
                        tempval = (tempval as number) * (currVal as number)
                        data.push(tempval);
                        break;

                }
            }
            return data;
        })

        if (axis === 0) {
            newData = utils.transposeArray(newData) as ArrayType2D
        }

        if (inplace) {
            this.$setValues(newData)
        } else {
            return new DataFrame(newData, {
                index: [...this.index],
                columns: [...this.columns],
                dtypes: [...this.dtypes],
                config: { ...this.config }
            })
        }
    }

    /**
     * Creates a deep copy of values while preserving NaN values
     * @param values The values to deep copy
     * @returns A deep copy of the values with NaN preserved
     */
    private $deepCopyValues(values: any): any {
        if (values === null || typeof values !== 'object') {
            return values;
        }
        if (values instanceof Date) {
            return new Date(values.getTime());
        }
        if (Array.isArray(values)) {
            return values.map(value => this.$deepCopyValues(value));
        }
        const result: any = {};
        for (const key in values) {
            if (Object.prototype.hasOwnProperty.call(values, key)) {
                result[key] = this.$deepCopyValues(values[key]);
            }
        }
        return result;
    }

    /**
     * Generate descriptive statistics for all numeric columns.
     * Descriptive statistics include those that summarize the central tendency,
     * dispersion and shape of a dataset’s distribution, excluding NaN values.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.describe().print()
     * ```
     */
    describe(): DataFrame {
        const numericColumnNames = this.columns.filter(name => this.$getColumnDtype(name) !== "string")
        const index = ["count", "mean", "std", "min", "median", "max", "variance"];
        const statsObject: any = {};
        for (let i = 0; i < numericColumnNames.length; i++) {
            const colName = numericColumnNames[i];
            const $count = (this.$getColumnData(colName) as Series).count();
            const $mean = mean(this.$getColumnData(colName, false) as number[]);
            const $std = std(this.$getColumnData(colName, false) as number[]);
            const $min = (this.$getColumnData(colName) as Series).min();
            const $median = median(this.$getColumnData(colName, false) as number[]);
            const $max = (this.$getColumnData(colName) as Series).max();
            const $variance = variance(this.$getColumnData(colName, false) as number[]);

            const stats = [$count, $mean, $std, $min, $median, $max, $variance];
            statsObject[colName] = stats;

        }

        const df = new DataFrame(statsObject, { index });
        return df
    }

    /**
     * Drops all rows or columns with missing values (NaN)
     * @param axis 0 or 1. If 0, drop columns with NaNs, if 1, drop rows with NaNs
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [NaN, NaN]], { columns: ['A', 'B']})
     * const df2 = df.dropna()
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [NaN, NaN]], { columns: ['A', 'B']})
     * df.dropna({ axis: 0, inplace: true }).print()
     * ```
    */
    dropNa(options?: { axis: 0 | 1, inplace?: boolean }): DataFrame
    dropNa(options?: { axis: 0 | 1, inplace?: boolean }): DataFrame | void {
        const { axis, inplace } = { axis: 1, inplace: false, ...options }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error("ParamError: Axis must be 0 or 1");
        }

        const newIndex: Array<number | string> = [];

        if (axis == 1) {
            const newData = [];

            const dfValues = this.values as ArrayType2D;
            for (let i = 0; i < dfValues.length; i++) {
                const values: ArrayType1D = dfValues[i];
                //@ts-ignore
                if (!values.includes(NaN) && !values.includes(undefined) && !values.includes(null)) {
                    newData.push(values);
                    newIndex.push(this.index[i])
                }
            }

            if (inplace) {
                this.$setValues(newData, false)
                this.$setIndex(newIndex)
            } else {
                return new DataFrame(
                    newData,
                    {
                        index: newIndex,
                        columns: [...this.columns],
                        dtypes: [...this.dtypes],
                        config: { ...this.config }
                    })
            }

        } else {
            const newColumnNames = []
            const newDtypes = []
            let dfValues: ArrayType2D = []

            if (this.config.isLowMemoryMode) {
                dfValues = utils.transposeArray(this.values) as ArrayType2D
            } else {
                dfValues = this.$dataIncolumnFormat as ArrayType2D
            }
            const tempColArr = []

            for (let i = 0; i < dfValues.length; i++) {
                const values: ArrayType1D = dfValues[i];
                if (!values.includes(NaN)) {
                    tempColArr.push(values);
                    newColumnNames.push(this.columns[i])
                    newDtypes.push(this.dtypes[i])
                }
            }

            const newData = utils.transposeArray(tempColArr) as ArrayType2D

            if (inplace) {
                this.$setValues(newData, false, false)
                this.$setColumnNames(newColumnNames)
                this.$setDtypes(newDtypes)
            } else {
                return new DataFrame(
                    newData,
                    {
                        index: [...this.index],
                        columns: newColumnNames,
                        dtypes: newDtypes,
                        config: { ...this.config }
                    })
            }
        }

    }

    /**
     * Adds a new column to the DataFrame. If column exists, then the column values is replaced. 
     * @param column The name of the column to add or replace. 
     * @param values An array of values to be inserted into the DataFrame. Must be the same length as the columns
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @param options.atIndex Column index to insert after. Defaults to the end of the columns.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.addColumn('C', [5, 6])
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.addColumn('C', [5, 6], { inplace: true }).print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.addColumn('C', [5, 6], { inplace: true, atIndex: 0 }).print()
     * ```
    */
    addColumn(
        column: string,
        values: Series | ArrayType1D,
        options?: { inplace?: boolean, atIndex?: number | string }
    ): DataFrame
    addColumn(
        column: string,
        values: Series | ArrayType1D,
        options?: { inplace?: boolean, atIndex?: number | string }
    ): DataFrame | void {
        let { inplace, atIndex } = { inplace: false, atIndex: this.columns.length, ...options };
        if (typeof atIndex === "string") {
            if (!(this.columns.includes(atIndex))) {
                throw new Error(`${atIndex} not a column`)
            }
            atIndex = this.columns.indexOf(atIndex)
        }

        if (!column) {
            throw new Error("ParamError: column must be specified")
        }

        if (!values) {
            throw new Error("ParamError: values must be specified")
        }

        const columnIndex = this.$columns.indexOf(column)

        if (columnIndex === -1) {
            let colunmValuesToAdd: ArrayType1D

            if (values instanceof Series) {
                colunmValuesToAdd = values.values as ArrayType1D
            } else if (Array.isArray(values)) {
                colunmValuesToAdd = values;
            } else {
                throw new Error("ParamError: specified value not supported. It must either be an Array or a Series of the same length")
            }

            if (colunmValuesToAdd.length !== this.shape[0]) {
                ErrorThrower.throwColumnLengthError(this, colunmValuesToAdd.length)
            }

            const newData = []
            const oldValues = this.$data
            for (let i = 0; i < oldValues.length; i++) {
                const innerArr = [...oldValues[i]] as ArrayType1D
                innerArr.splice(atIndex, 0, colunmValuesToAdd[i])
                newData.push(innerArr)
            }

            if (inplace) {
                this.$setValues(newData, true, false)
                let columns = [...this.columns]
                columns.splice(atIndex, 0, column)
                this.$setColumnNames(columns)
                this.$setInternalColumnDataProperty(column);

            } else {
                let columns = [...this.columns]
                columns.splice(atIndex, 0, column)

                const df = new DataFrame(newData, {
                    index: [...this.index],
                    columns: columns,
                    dtypes: [...this.dtypes, utils.inferDtype(colunmValuesToAdd)[0]],
                    config: { ...this.$config }
                })
                return df
            }
        } else {
            this.$setColumnData(column, values);
        }

    }

    /**
     * Makes a deep copy of a DataFrame.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.copy()
     * df2.print()
     * ```
     */
    copy(): DataFrame {
        let df = new DataFrame([...this.$data], {
            columns: [...this.columns],
            index: [...this.index],
            dtypes: [...this.dtypes],
            config: { ...this.$config }
        });
        return df;
    }

    /**
     * Return a boolean, same-sized object indicating where elements are empty (NaN, undefined, null).
     * NaN, undefined and null values gets mapped to true, and everything else gets mapped to false.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.isNa().print()
     * ```
    */
    isNa(): DataFrame {
        const newData = []
        for (let i = 0; i < this.values.length; i++) {
            const valueArr = this.values[i] as ArrayType1D
            const tempData = valueArr.map((value) => {
                if (utils.isEmpty(value)) {
                    return true;
                } else {
                    return false;
                }
            })
            newData.push(tempData)
        }

        const df = new DataFrame(newData,
            {
                index: [...this.index],
                columns: [...this.columns],
                config: { ...this.config }
            });
        return df;
    }

    /**
    * Replace all empty elements with a specified value. Replace params expect columns array to map to values array.
    * @param values The list of values to use for replacement.
    * @param options.columns  The list of column names to be replaced.
    * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false.
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4], [NaN, NaN]], { columns: ['A', 'B']})
    * const df2 = df.fillNa(-99)
    * df2.print()
    * ```
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4], [NaN, NaN]], { columns: ['A', 'B']})
    * df.fillNa(-99, { inplace: true }).print()
    * ```
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4], [NaN, NaN]], { columns: ['A', 'B']})
    * df.fillNa(-99, { columns: ["A"], inplace: true }).print()
    * ```
    * 
    */
    fillNa(
        values: number | string | boolean | ArrayType1D,
        options?:
            {
                columns?: Array<string>,
                inplace?: boolean
            }
    ): DataFrame
    fillNa(
        values: number | string | boolean | ArrayType1D,
        options?:
            {
                columns?: Array<string>,
                inplace?: boolean
            }
    ): DataFrame | void {
        let { columns, inplace } = { inplace: false, ...options }

        if (!values && typeof values !== "boolean" && typeof values !== "number" && typeof values !== "string") {
            throw Error('ParamError: value must be specified');
        }

        if (Array.isArray(values)) {
            if (!Array.isArray(columns)) {
                throw Error('ParamError: value is an array, hence columns must also be an array of same length');
            }

            if (values.length !== columns.length) {
                throw Error('ParamError: specified column and values must have the same length');
            }

            columns.forEach((col) => {
                if (!this.columns.includes(col)) {
                    throw Error(
                        `ValueError: Specified column "${col}" must be one of ${this.columns}`
                    );
                }
            });
        }

        const newData = []
        const oldValues = [...this.values]

        if (!columns) {
            //Fill all columns
            for (let i = 0; i < oldValues.length; i++) {
                let valueArr = [];
                if (inplace) {
                    valueArr = this.$deepCopyValues(oldValues[i]) as ArrayType1D;
                } else {
                    valueArr = [...oldValues[i] as ArrayType1D];
                }

                const tempArr = valueArr.map((innerVal) => {
                    if (utils.isEmpty(innerVal)) {
                        const replaceWith = Array.isArray(values) ? values[i] : values
                        return replaceWith
                    } else {
                        return innerVal
                    }
                })
                newData.push(tempArr)
            }

        } else {
            //Fill specific columns
            for (let i = 0; i < oldValues.length; i++) {
                let valueArr = [];
                if (inplace) {
                    valueArr = this.$deepCopyValues(oldValues[i]) as ArrayType1D;
                } else {
                    valueArr = [...oldValues[i] as ArrayType1D];
                }

                for (let j = 0; j < columns.length; j++) { 
                    const columnIndex = this.columns.indexOf(columns[j])
                    const replaceWith = Array.isArray(values) ? values[j] : values
                    valueArr[columnIndex] = utils.isEmpty(valueArr[columnIndex]) ? replaceWith : valueArr[columnIndex]
                }
                newData.push(valueArr)
            }
        }

        if (inplace) {
            this.$setValues(newData as ArrayType2D)
        } else {
            const df = new DataFrame(newData,
                {
                    index: [...this.index],
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                });
            return df;
        }
    }

    /**
    * Drop specified columns or rows.
    * @param options.columns Array of column names to drop.
    * @param options.index Array of index to drop.
    * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false.
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
    * const df2 = df.drop({ columns: ['A'] })
    * df2.print()
    * ```
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
    * df.drop({ index: [0], inplace: true }).print()
    * ```
    */
    drop(options?:
        {
            columns?: string | Array<string>,
            index?: Array<string | number>,
            inplace?: boolean
        }
    ): DataFrame
    drop(options?:
        {
            columns?: string | Array<string>,
            index?: Array<string | number>,
            inplace?: boolean
        }
    ): DataFrame | void {
        let { columns, index, inplace } = { inplace: false, ...options }

        if (!columns && !index) {
            throw Error('ParamError: Must specify one of columns or index');
        }

        if (columns && index) {
            throw Error('ParamError: Can only specify one of columns or index');
        }

        if (columns) {
            const columnIndices: Array<number> = []

            if (typeof columns === "string") {
                columnIndices.push(this.columns.indexOf(columns))
            } else if (Array.isArray(columns)) {
                for (let column of columns) {
                    if (this.columns.indexOf(column) === -1) {
                        throw Error(`ParamError: specified column "${column}" not found in columns`);
                    }
                    columnIndices.push(this.columns.indexOf(column))
                }

            } else {
                throw Error('ParamError: columns must be an array of column names or a string of column name');
            }

            let newRowData: ArrayType2D = []
            let newColumnNames = []
            let newDtypes = []

            for (let i = 0; i < this.values.length; i++) {
                const tempInnerArr = []
                const innerArr = this.values[i] as ArrayType1D
                for (let j = 0; j < innerArr.length; j++) {
                    if (!(columnIndices.includes(j))) {
                        tempInnerArr.push(innerArr[j])
                    }
                }
                newRowData.push(tempInnerArr)
            }

            for (let i = 0; i < this.columns.length; i++) {
                const element = this.columns[i]
                if (!(columns.includes(element))) {
                    newColumnNames.push(element)
                    newDtypes.push(this.dtypes[i])
                }
            }

            if (inplace) {
                this.$setValues(newRowData, true, false)
                this.$setColumnNames(newColumnNames)
            } else {
                const df = new DataFrame(newRowData,
                    {
                        index: [...this.index],
                        columns: newColumnNames,
                        dtypes: newDtypes,
                        config: { ...this.config }
                    });
                return df;
            }

        }

        if (index) {
            const rowIndices: Array<number> = []

            if (typeof index === "string" || typeof index === "number" || typeof index === "boolean") {
                rowIndices.push(this.index.indexOf(index))
            } else if (Array.isArray(index)) {
                for (let indx of index) {
                    if (this.index.indexOf(indx) === -1) {
                        throw Error(`ParamError: specified index "${indx}" not found in indices`);
                    }
                    rowIndices.push(this.index.indexOf(indx));
                }
            } else {
                throw Error('ParamError: index must be an array of indices or a scalar index');
            }

            let newRowData: ArrayType2D = []
            let newIndex = []

            for (let i = 0; i < this.values.length; i++) {
                const innerArr = this.values[i] as ArrayType1D
                if (!(rowIndices.includes(i))) {
                    newRowData.push(innerArr)
                }
            }

            for (let i = 0; i < this.index.length; i++) {
                const indx = this.index[i]
                if (!(index.includes(indx))) {
                    newIndex.push(indx)
                }
            }

            if (inplace) {
                this.$setValues(newRowData, false)
                this.$setIndex(newIndex)
            } else {
                const df = new DataFrame(newRowData,
                    {
                        index: newIndex,
                        columns: [...this.columns],
                        dtypes: [...this.dtypes],
                        config: { ...this.config }
                    });
                return df;
            }
        }

    }

    /**
    * Sorts a Dataframe by a specified column values
    * @param column Column name to sort by.
    * @param options.ascending Whether to sort values in ascending order or not. Defaults to true.
    * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false.
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
    * const df2 = df.sortBy('A')
    * df2.print()
    * ```
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
    * df.sortBy('A', { ascending: false, inplace: true }).print()
    * ```
    */
    sortValues(
        column: string,
        options?:
            {
                ascending?: boolean
                inplace?: boolean
            }
    ): DataFrame
    sortValues(
        column: string,
        options?:
            {
                ascending?: boolean
                inplace?: boolean
            }
    ): DataFrame | void {
        const { ascending, inplace } = { ascending: true, inplace: false, ...options }

        if (!column) {
            throw Error(`ParamError: must specify a column to sort by`);
        }

        if (this.columns.indexOf(column) === -1) {
            throw Error(`ParamError: specified column "${column}" not found in columns`);
        }

        const columnValues = this.$getColumnData(column, false) as ArrayType1D
        const index = [...this.index]

        const objToSort = columnValues.map((value, i) => {
            return { index: index[i], value }
        })

        const sortedObjectArr = utils.sortObj(objToSort, ascending)
        const sortedIndex = sortedObjectArr.map(obj => obj.index)

        const newDf = _loc({ ndFrame: this, rows: sortedIndex }) as DataFrame

        if (inplace) {
            this.$setValues(newDf.values)
            this.$setIndex(newDf.index)
        } else {
            return newDf
        }

    }

    /**
       * Sets the index of the DataFrame to the specified value.
       * @param options.index An array of index values to set
       * @param options.column A column name whose values set in place of the index
       * @param options.drop Whether to drop the column whose index was set. Defaults to false
       * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
       * @example
       * ```
       * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
       * const df2 = df.setIndex({ index: ['a', 'b'] })
       * df2.print()
       * ```
       * 
       * @example
       * ```
       * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
       * df.setIndex({ column: "A", inplace: true })
       * df.print()
       * ```
    */
    setIndex(
        options:
            {
                index?: Array<number | string | (number | string)>,
                column?: string,
                drop?: boolean,
                inplace?: boolean
            }
    ): DataFrame
    setIndex(
        options:
            {
                index?: Array<number | string | (number | string)>,
                column?: string,
                drop?: boolean,
                inplace?: boolean
            }
    ): DataFrame | void {
        const { index, column, drop, inplace } = { drop: false, inplace: false, ...options }

        if (!index && !column) {
            throw new Error("ParamError: must specify either index or column")
        }

        let newIndex: Array<string | number> = [];

        if (index) {
            if (!Array.isArray(index)) {
                throw Error(`ParamError: index must be an array`);
            }

            if (index.length !== this.values.length) {
                throw Error(`ParamError: index must be the same length as the number of rows`);
            }
            newIndex = index;
        }

        if (column) {
            if (this.columns.indexOf(column) === -1) {
                throw Error(`ParamError: column not found in column names`);
            }

            newIndex = this.$getColumnData(column, false) as Array<string | number>
        }

        if (drop) {
            const dfDropped = this.drop({ columns: [column as string] })

            const newData = dfDropped?.values as ArrayType2D
            const newColumns = dfDropped?.columns
            const newDtypes = dfDropped?.dtypes

            if (inplace) {
                this.$setValues(newData, true, false)
                this.$setIndex(newIndex)
                this.$setColumnNames(newColumns)
            } else {
                const df = new DataFrame(newData,
                    {
                        index: newIndex,
                        columns: newColumns,
                        dtypes: newDtypes,
                        config: { ...this.config }
                    });
                return df;
            }
        } else {
            if (inplace) {
                this.$setIndex(newIndex)
            } else {
                const df = new DataFrame(this.values,
                    {
                        index: newIndex,
                        columns: [...this.columns],
                        dtypes: [...this.dtypes],
                        config: { ...this.config }
                    });
                return df;
            }
        }

    }

    /**
       * Resets the index of the DataFrame to default.
       * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
       * @example
       * ```
       * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
       * df.resetIndex({ inplace: true })
       * df.print()
       * ```
    */
    resetIndex(options?: { inplace?: boolean }): DataFrame
    resetIndex(options?: { inplace?: boolean }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        if (inplace) {
            this.$resetIndex()
        } else {
            const df = new DataFrame(this.values,
                {
                    index: this.index.map((_, i) => i),
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                });
            return df;
        }

    }

    /**
     * Apply a function along an axis of the DataFrame. To apply a function element-wise, use `applyMap`.
     * Objects passed to the function are Series values whose 
     * index is either the DataFrame’s index (axis=0) or the DataFrame’s columns (axis=1)
     * @param callable Function to apply to each column or row.
     * @param options.axis 0 or 1. If 0, apply "callable" column-wise, else apply row-wise
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.apply(Math.sqrt, { axis: 0 })
     * df2.print()
     * ```
    */
    apply(callable: any, options?: { axis?: 0 | 1 }): DataFrame | Series {
        const { axis } = { axis: 1, ...options }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error(`ParamError: axis must be 0 or 1`);
        }

        const valuesForFunc = this.$getDataByAxisWithMissingValuesRemoved(axis)

        const result = valuesForFunc.map(row => {
            return callable(row)
        })

        if (axis === 0) {
            if (utils.is1DArray(result)) {
                return new Series(result, {
                    index: [...this.columns]
                })
            } else {
                return new DataFrame(result, {
                    index: [...this.columns],
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                })
            }
        } else {
            if (utils.is1DArray(result)) {
                return new Series(result, {
                    index: [...this.index]
                })
            } else {
                return new DataFrame(result, {
                    index: [...this.index],
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                })
            }
        }

    }

    /**
     * Apply a function to a Dataframe values element-wise.
     * @param callable Function to apply to each column or row
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * function square(x) { return x * x }
     * const df2 = df.applyMap(square)
     * df2.print()
     * ```
    */
    applyMap(callable: any, options?: { inplace?: boolean }): DataFrame
    applyMap(callable: any, options?: { inplace?: boolean }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        const newData = (this.values as ArrayType2D).map((row) => {
            const tempData = row.map((val) => {
                return callable(val);
            });
            return tempData;
        });

        if (inplace) {
            this.$setValues(newData);
        } else {
            return new DataFrame(newData,
                {
                    index: [...this.index],
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                });
        }
    }

    /**
     * Returns the specified column data as a Series object.
     * @param column The name of the column to return
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const sf = df.column('A')
     * sf.print()
     * ```
     * 
    */
    column(column: string): Series {
        return this.$getColumnData(column) as Series
    }

    /**
     * Return a subset of the DataFrame based on the column dtypes.
     * @param include An array of dtypes or strings to be included.
     * @example
     * ```
     * const df = new DataFrame([[1, 2.1, "Dog"], [3, 4.3, "Cat"]], { columns: ['A', 'B', 'C']})
     * const df2 = df.selectDtypes(['float32'])
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2.1, "Dog"], [3, 4.3, "Cat"]], { columns: ['A', 'B', 'C']})
     * const df2 = df.selectDtypes(['float32', 'int32'])
     * df2.print()
     * ```
     * 
    */
    selectDtypes(include: Array<string>): DataFrame {
        const supportedDtypes = ["float32", "int32", "string", "boolean", 'undefined']

        if (Array.isArray(include) === false) {
            throw Error(`ParamError: include must be an array`);
        }

        include.forEach(dtype => {
            if (supportedDtypes.indexOf(dtype) === -1) {
                throw Error(`ParamError: include must be an array of valid dtypes`);
            }
        })
        const newColumnNames = []

        for (let i = 0; i < this.dtypes.length; i++) {
            if (include.includes(this.dtypes[i])) {
                newColumnNames.push(this.columns[i])
            }
        }
        return this.loc({ columns: newColumnNames })

    }

    /**
     * Returns the transpose of the DataFrame.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.transpose()
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.transpose({ inplace: true })
     * df.print()
     * ```
     **/
    transpose(options?: { inplace?: boolean }): DataFrame
    transpose(options?: { inplace?: boolean }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }
        const newData = utils.transposeArray(this.values)
        const newColNames = [...this.index.map(i => i.toString())]

        if (inplace) {
            this.$setValues(newData, false, false)
            this.$setIndex([...this.columns])
            this.$setColumnNames(newColNames)
        } else {
            return new DataFrame(newData, {
                index: [...this.columns],
                columns: newColNames,
                config: { ...this.config }
            })
        }
    }

    /**
     * Returns the Transpose of the DataFrame. Similar to `transpose`.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.T()
     * df2.print()
     * ```
    **/
    get T(): DataFrame {
        const newData = utils.transposeArray(this.values)
        return new DataFrame(newData, {
            index: [...this.columns],
            columns: [...this.index.map(i => i.toString())],
            config: { ...this.config }
        })
    }

    /**
      * Replace all occurence of a value with a new value.
      * @param oldValue The value you want to replace
      * @param newValue The new value you want to replace the old value with
      * @param options.columns An array of column names you want to replace. If not provided replace accross all columns.
      * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
      * 
      * @example
      * ```
      * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
      * const df2 = df.replace(2, 5)    
      * df.print()
      * ```
      * 
      * @example
      * ```
      * const df = new DataFrame([[1, 2], [2, 20]], { columns: ['A', 'B']})
      * const df2 = df.replace(2, 5, { columns: ['A'] })
      * df2.print()
      * ```
    */
    replace(
        oldValue: number | string | boolean,
        newValue: number | string | boolean,
        options?: {
            columns?: Array<string>
            inplace?: boolean
        }
    ): DataFrame
    replace(
        oldValue: number | string | boolean,
        newValue: number | string | boolean,
        options?: {
            columns?: Array<string>
            inplace?: boolean
        }
    ): DataFrame | void {
        const { columns, inplace } = { inplace: false, ...options }

        if (typeof oldValue === 'number' && isNaN(oldValue)) {
            throw Error(`Params Error: Param 'oldValue' does not support NaN. Use DataFrame.fillNa() instead.`);
        }

        if (!oldValue && typeof oldValue !== 'boolean' && typeof oldValue !== 'number' && typeof oldValue !== 'string') {
            throw Error(`Params Error: Must specify param 'oldValue' to replace`);
        }

        if (!newValue && typeof newValue !== 'boolean' && typeof newValue !== 'number' && typeof newValue !== 'string') {
            throw Error(`Params Error: Must specify param 'newValue' to replace with`);
        }

        let newData: ArrayType2D = []

        if (columns) {
            if (!Array.isArray(columns)) {
                throw Error(`Params Error: column must be an array of column(s)`);
            }
            const columnIndex: Array<number> = []

            columns.forEach(column => {
                const _indx = this.columns.indexOf(column)
                if (_indx === -1) {
                    throw Error(`Params Error: column not found in columns`);
                }
                columnIndex.push(_indx)
            })

            newData = (this.values as ArrayType2D).map(([...row]) => {
                for (const colIndx of columnIndex) {
                    if (row[colIndx] === oldValue) {
                        row[colIndx] = newValue;
                    }
                }
                return row;
            })
        } else {
            newData = (this.values as ArrayType2D).map(([...row]) => {
                return row.map((cell => {
                    if (cell === oldValue) {
                        return newValue
                    } else {
                        return cell
                    }
                }))
            })
        }

        if (inplace) {
            this.$setValues(newData)
        } else {
            return new DataFrame(newData, {
                index: [...this.index],
                columns: [...this.columns],
                dtypes: [...this.dtypes],
                config: { ...this.config }
            })
        }
    }



    /**
     * Cast the values of a column to specified data type
     * @param column The name of the column to cast
     * @param dtype Data type to cast to. One of [float32, int32, string, boolean]
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2.2], [3, 4.3]], { columns: ['A', 'B']})
     * const df2 = df.asType('B', 'int32')
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2.2], [3, 4.3]], { columns: ['A', 'B']})
     * df.asType('B', 'int32', { inplace: true })
     * df.print()
     * ```
     */
    asType(
        column: string,
        dtype: "float32" | "int32" | "string" | "boolean",
        options?: { inplace?: boolean }
    ): DataFrame
    asType(
        column: string,
        dtype: "float32" | "int32" | "string" | "boolean",
        options?: { inplace?: boolean }
    ): DataFrame | void {
        const { inplace } = { inplace: false, ...options }
        const columnIndex = this.columns.indexOf(column)

        if (columnIndex === -1) {
            throw Error(`Params Error: column not found in columns`);
        }

        if (!(DATA_TYPES.includes(dtype))) {
            throw Error(`dtype ${dtype} not supported. dtype must be one of ${DATA_TYPES}`);
        }

        const data = this.values as ArrayType2D

        const newData = data.map((row) => {
            if (dtype === "float32") {
                row[columnIndex] = Number(row[columnIndex])
                return row
            } else if (dtype === "int32") {
                row[columnIndex] = parseInt(row[columnIndex] as any)
                return row
            } else if (dtype === "string") {
                row[columnIndex] = row[columnIndex].toString()
                return row
            } else if (dtype === "boolean") {
                row[columnIndex] = Boolean(row[columnIndex])
                return row
            }
        })

        if (inplace) {
            this.$setValues(newData as any)
        } else {
            const newDtypes = [...this.dtypes]
            newDtypes[columnIndex] = dtype

            return new DataFrame(newData, {
                index: [...this.index],
                columns: [...this.columns],
                dtypes: newDtypes,
                config: { ...this.config }
            })
        }
    }

    /**
     * Return the number of unique elements in a column, across the specified axis.
     * To get the values use `.unique()` instead. 
     * @param axis The axis to count unique elements across. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * df.nunique().print()
     * ```
     * 
    */
    nUnique(axis: 0 | 1 = 1): Series {
        if ([0, 1].indexOf(axis) === -1) {
            throw Error(`ParamError: axis must be 0 or 1`);
        }
        const data = this.$getDataArraysByAxis(axis)
        const newData = data.map(row => new Set(row).size)

        if (axis === 0) {
            return new Series(newData, {
                index: [...this.columns],
                dtypes: ["int32"]
            })
        } else {
            return new Series(newData, {
                index: [...this.index],
                dtypes: ["int32"]
            })
        }
    }

    /**
     * Renames a column or index to specified value. 
     * @param mapper An object that maps each column or index in the DataFrame to a new value
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @param options.axis The axis to perform the operation on. Defaults to 1
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const df2 = df.rename({ A: 'a', B: 'b' })
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * df.rename({ A: 'a', B: 'b' }, { inplace: true })
     * df.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * df.rename({ 0: 'a', 1: 'b' }, { axis: 0, inplace: true})
     * df.print()
     * ```
     * 
     */
    rename(
        mapper: {
            [index: string | number]: string | number
        },
        options?: {
            axis?: 0 | 1
            inplace?: boolean
        }
    ): DataFrame
    rename(
        mapper: {
            [index: string | number]: string | number
        },
        options?: {
            axis?: 0 | 1
            inplace?: boolean
        }
    ): DataFrame | void {
        const { axis, inplace } = { axis: 1, inplace: false, ...options }

        if ([0, 1].indexOf(axis) === -1) {
            throw Error(`ParamError: axis must be 0 or 1`);
        }

        if (axis === 1) {
            const colsAdded: string[] = [];
            const newColumns = this.columns.map(col => {
                if (mapper[col] !== undefined) {
                    const newCol = `${mapper[col]}`;
                    colsAdded.push(newCol);
                    return newCol;
                } else {
                    return col
                }
            })

            if (inplace) {
                this.$setColumnNames(newColumns)
                for (const col of colsAdded) {
                    this.$setInternalColumnDataProperty(col);
                }
            } else {
                return new DataFrame([...this.values], {
                    index: [...this.index],
                    columns: newColumns,
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                })
            }
        } else {
            const newIndex = this.index.map(col => {
                if (mapper[col] !== undefined) {
                    return mapper[col]
                } else {
                    return col
                }
            })

            if (inplace) {
                this.$setIndex(newIndex)
            } else {
                return new DataFrame([...this.values], {
                    index: newIndex,
                    columns: [...this.columns],
                    dtypes: [...this.dtypes],
                    config: { ...this.config }
                })
            }
        }

    }

    /**
    * Sorts the Dataframe by the index.
    * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
    * @param options.ascending Whether to sort values in ascending order or not. Defaults to true
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
    * const df2 = df.sortIndex()
    * df2.print()
    * ```
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
    * df.sortIndex({ inplace: true })
    * df.print()
    * ```
    * 
    * @example
    * ```
    * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
    * df.sortIndex({ ascending: false, inplace: true })
    * df.print()
    * ```
    */
    sortIndex(options?:
        {
            inplace?: boolean
            ascending?: boolean
        }
    ): DataFrame
    sortIndex(options?:
        {
            inplace?: boolean
            ascending?: boolean
        }
    ): DataFrame | void {
        const { ascending, inplace } = { ascending: true, inplace: false, ...options }

        const indexPosition = utils.range(0, this.index.length - 1)
        const index = [...this.index]

        const objToSort = index.map((idx, i) => {
            return { index: indexPosition[i], value: idx }
        })

        const sortedObjectArr = utils.sortObj(objToSort, ascending)
        let sortedIndex = sortedObjectArr.map((obj) => obj.index);
        const newData = sortedIndex.map(i => (this.values as ArrayType2D)[i as number])
        sortedIndex = sortedIndex.map((i) => index[i as number]);

        if (inplace) {
            this.$setValues(newData)
            this.$setIndex(sortedIndex)
        } else {
            return new DataFrame(newData, {
                index: sortedIndex,
                columns: [...this.columns],
                dtypes: [...this.dtypes],
                config: { ...this.config }
            })
        }

    }

    /**
     * Add new rows at the end of the DataFrame.
     * @param newValues Array, Series or DataFrame to append to the DataFrame 
     * @param index The new index value(s) to append to the Series. Must contain the same number of values as `newValues`
     * as they map `1 - 1`. 
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const values = [7, 8]
     * const index = ['a', 'b']
     * const df2 = df.append(values, index)
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const values = new Series([7, 8, 9, 10])
     * const index = ['a', 'b', 'c', 'd']
     * const df2 = df.append(values, index)
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const values = new DataFrame([[7, 8], [9, 10]], { columns: ['C', 'D'] })
     * const index = ['a', 'b']
     * const df2 = df.append(values, index)
     * df2.print()
     * ```
     */
    append(
        newValues: ArrayType1D | ArrayType2D | Series | DataFrame,
        index: Array<number | string> | number | string,
        options?: {
            inplace?: boolean,
        }
    ): DataFrame
    append(
        newValues: ArrayType1D | ArrayType2D | Series | DataFrame,
        index: Array<number | string> | number | string,
        options?: {
            inplace?: boolean,
        }
    ): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        if (!newValues) {
            throw Error(`ParamError: newValues must be a Series, DataFrame or Array`);
        }

        if (!index) {
            throw Error(`ParamError: index must be specified`);
        }

        let rowsToAdd = []
        if (newValues instanceof Series) {

            if (newValues.values.length !== this.shape[1]) {
                throw Error(`ValueError: length of newValues must be the same as the number of columns.`);
            }
            rowsToAdd = [newValues.values]

        } else if (newValues instanceof DataFrame) {

            if (newValues.shape[1] !== this.shape[1]) {
                throw Error(`ValueError: length of newValues must be the same as the number of columns.`);
            }
            rowsToAdd = newValues.values

        } else if (Array.isArray(newValues)) {

            if (utils.is1DArray(newValues)) {
                rowsToAdd = [newValues]
            } else {
                rowsToAdd = newValues
            }

            if ((rowsToAdd[0] as any).length !== this.shape[1]) {
                throw Error(`ValueError: length of newValues must be the same as the number of columns.`);
            }

        } else {
            throw Error(`ValueError: newValues must be a Series, DataFrame or Array`);
        }


        let indexInArrFormat: Array<number | string> = []
        if (!Array.isArray(index)) {
            indexInArrFormat = [index]
        } else {
            indexInArrFormat = index
        }

        if (rowsToAdd.length !== indexInArrFormat.length) {
            throw Error(`ParamError: index must contain the same number of values as newValues`);
        }

        const newData = [...this.values] as ArrayType2D
        const newIndex = [...this.index]

        rowsToAdd.forEach((row, i) => {
            newData.push(row as ArrayType1D)
            newIndex.push(indexInArrFormat[i])
        })

        if (inplace) {
            this.$setValues(newData)
            this.$setIndex(newIndex)
        } else {
            return new DataFrame(newData, {
                index: newIndex,
                columns: [...this.columns],
                dtypes: [...this.dtypes],
                config: { ...this.config }
            })
        }

    }

    /**
     * Queries the DataFrame for rows that meet the boolean criteria. This is just a wrapper for the `iloc` method.
     * @param condition An array of boolean mask, one for each row in the DataFrame. Rows where the value are true will be returned.
     * @param options.inplace Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const df2 = df.query([true, false, true, true])
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const df2 = df.query(df["A"].gt(2))
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * const df2 = df.query(df["A"].gt(2).and(df["B"].lt(5)))
     * df2.print()
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4], [1, 2], [5, 6]], { columns: ['A', 'B'] })
     * df.query(df["A"].gt(2), { inplace: true })
     * df.print()
     * ```
    **/
    query(condition: Series | Array<boolean>, options?: { inplace?: boolean }): DataFrame
    query(condition: Series | Array<boolean>, options?: { inplace?: boolean }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        if (!condition) {
            throw new Error("ParamError: condition must be specified");
        }

        const result = _iloc({
            ndFrame: this,
            rows: condition,
        }) as DataFrame

        if (inplace) {
            this.$setValues(result.values, false, false)
            this.$setIndex(result.index)
        } else {
            return result
        }

    }

    /**
     * Returns the data types for each column as a Series.
     * @example
     * ```
     * const df = new DataFrame([[1, 2.1, "Dog"], [3, 4.3, "Cat"]], { columns: ['A', 'B', 'C'] })
     * df.ctypes().print()
     * ```
     */
    get ctypes(): Series {
        return new Series(this.dtypes, { index: this.columns })
    }

    /**
     * One-hot encode specified columns in the DataFrame. If columns are not specified, all columns of string dtype will be encoded.
     * @param options Options for the operation. The following options are available:
     * - `columns`: A single column name or an array of column names to encode. Defaults to all columns of dtype string.
     * - `prefix`: Prefix to add to the column names. Defaults to unique labels.
     * - `prefixSeparator`: Separator to use for the prefix. Defaults to '_'.
     * - `inplace`: Boolean indicating whether to perform the operation inplace or not. Defaults to false
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.getDummies()
     * ```
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.getDummies({ columns: ['A'] })
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.getDummies({ prefix: 'cat' })
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.getDummies({ prefix: 'cat', prefixSeparator: '_' })
     * ```
     * 
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * const df2 = df.getDummies({ inplace: true })
     * ```
     */
    getDummies(options?: {
        columns?: string | Array<string>,
        prefix?: string | Array<string>,
        prefixSeparator?: string | Array<string>,
        inplace?: boolean
    }): DataFrame
    getDummies(options?: {
        columns?: string | Array<string>,
        prefix?: string | Array<string>,
        prefixSeparator?: string | Array<string>,
        inplace?: boolean
    }): DataFrame | void {
        const { inplace } = { inplace: false, ...options }

        const encodedDF = dummyEncode(this, options)
        if (inplace) {
            this.$setValues(encodedDF.values, false, false)
            this.$setColumnNames(encodedDF.columns)
        } else {
            return encodedDF
        }

    }

    /**
     * Groupby
     * @params col a list of column
     * @returns Groupby
     * @example
     * let data = [ [ 1, 2, 3 ], [ 4, 5, 6 ], [ 20, 30, 40 ], [ 39, 89, 78 ] ];
     * let cols = [ "A", "B", "C" ];
     * let df = new dfd.DataFrame(data, { columns: cols });
     * let groupDf = df.groupby([ "A" ]);
     */
    groupby(col: Array<string>): Groupby {
        const columns = this.columns
        const colIndex = col.map((val) => columns.indexOf(val))
        const colDtype = this.dtypes

        return new Groupby(
            col,
            this.values as ArrayType2D,
            columns,
            colDtype,
            colIndex
        ).group()

    }

    /**
     * Access a single value for a row/column pair by integer position.
     * Similar to {@link iloc}, in that both provide integer-based lookups. 
     * Use iat if you only need to get or set a single value in a DataFrame.
     * @param row Row index of the value to access.
     * @param column Column index of the value to access.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.iat(0, 0) // 1
     * df.iat(0, 1) // 2
     * df.iat(1, 0) // 3
     * ```
    */
    iat(row: number, column: number): string | number | boolean | undefined {
        if (typeof row === 'string' || typeof column === 'string') {
            throw new Error('ParamError: row and column index must be an integer. Use .at to get a row or column by label.')
        }

        return (this.values as ArrayType2D)[row][column]
    }

    /**
     * Access a single value for a row/column label pair.
     * Similar to {@link loc}, in that both provide label-based lookups. 
     * Use at if you only need to get or set a single value in a DataFrame.
     * @param row Row index of the value to access.
     * @param column Column label of the value to access.
     * @example
     * ```
     * const df = new DataFrame([[1, 2], [3, 4]], { columns: ['A', 'B']})
     * df.at(0,'A') // 1
     * df.at(1, 'A') // 3
     * df.at(1, 'B') // 4
     * ```
    */
    at(row: string | number, column: string): string | number | boolean | undefined {
        if (typeof column !== 'string') {
            throw new Error('ParamError: column index must be a string. Use .iat to get a row or column by index.')
        }
        return (this.values as ArrayType2D)[this.index.indexOf(row)][this.columns.indexOf(column)]
    }

    /**
     * Exposes functions for creating charts from a DataFrame. 
     * Charts are created using the Plotly.js library, so all Plotly's configuration parameters are available.
     * @param divId name of the HTML Div to render the chart in.
    */
    plot(divId: string): IPlotlyLib {
        //TODO: Add support for check plot library to use. So we can support other plot library like d3, vega, etc
        if (utils.isBrowserEnv()) {
            const plt = new PlotlyLib(this, divId);
            return plt;
        } else {
            throw new Error("Not supported in NodeJS");
        }
    }
}