Retail AI Insights: Customer Intelligence with Hybrid ML

Project Overview

This MVP was developed during a 6-hour Critical Challenge in a Corporate Bootcamp. The goal was to transform raw transactional data (50,000 records) into a strategic decision-making tool for the Retail sector. We implemented a Hybrid Machine Learning engine that automatically segments customers and uncovers the business rules behind each group.

Methodology and Architecture

The project follows a clean architecture, separating data logic from the user interface to ensure scalability and maintainability:

1. RFM Engineering: Data processing to calculate Recency, Frequency, and Monetary value per customer.
2. Unsupervised Learning (K-Means): Discovery of 4 natural behavior clusters.
3. Explainable AI (Decision Tree): Supervised model to extract logical rules and validate segmentation with a 70/30 data split.
4. Performance Optimization: Data ingestion via CSV with caching (@st.cache_data) for millisecond response times.

Impact Results (KPIs)

• Model Accuracy: 90.1% in classifying new customers.
• Analysis Scope: 5,708 unique customers identified.
• Total Sales Analyzed: $7,492,725.
• Global Average Ticket: $1,312.67.
• Critical Segmentation: Identification of 1,017 VIP customers sustaining business profitability.

Technologies Used

• Language: Python 3.9+
• Data Analysis: Pandas, Numpy
• Machine Learning: Scikit-Learn (KMeans, DecisionTreeClassifier)
• Visualization: Plotly Express, Matplotlib
• Interface: Streamlit (corporate tab-based layout)

Repository Structure

IA-BOOTCAMP/
├── .venv/               # Entorno virtual
├── app.py               # Interfaz de usuario y Dashboard
├── logic.py             # Motor de Machine Learning y RFM
├── Data_clean_model.csv # Dataset optimizado de 50k registros
├── requirements.txt     # Dependencias del proyecto
└── README.md            # Documentación estratégica

Sprint Execution (Scrum)

• 11:00 AM: Data ingestion and outlier cleaning.
• 01:00 PM: Development of RFM engine and clustering.
• 03:00 PM: Training of supervised model and metric validation.
• 04:30 PM: UI/UX refinement and load optimization.

How to Run the Project

1. Clone the repository:

   git clone <url-del-repositorio>
   cd IA-BOOTCAMP

Install dependencies:

   pip install -r requirements.txt

Run the application:

   streamlit run app.py

Contributions Contributions are welcome. Please open an issue to discuss major changes before submitting a pull request.

License This project was developed for educational purposes during a corporate bootcamp.

If you found this project useful, consider giving it a star on GitHub.