[Query] Missing Implementation in DecisionListLearner in learning.py

[ayush113]

Hello !

This issue is about the DecisionListLearner function within learning.py

I noticed that the implementation of find_examples and passes function just raises NotImplementedError. I was wondering if I could take this up and wanted to find out about any requirements/information about this.

@MrDupin @ad71

Thank You !

[JaakTepandi]

Hello!
Seems the pull request is open on #998.

[dmeoli]

Implemented in 5b97739. DecisionListLearner previously had unimplemented find_examples/passes (both raised NotImplementedError) and called set(dataset.examples) on list-valued (unhashable) examples, so it couldn't run.

It now implements the Figure 18.11 algorithm: a test is a conjunction of (attribute, value) literals; find_examples returns the smallest test that selects a non-empty subset of examples all sharing one outcome; learning appends (test, outcome) and recurses on the remaining examples, ending in an empty catch-all test; predict returns the outcome of the first satisfied test. Fixed in both learning.py (closure form) and learning4e.py (rewritten as a proper class, replacing the broken self.predict.decision_list bound-method-attribute pattern). Verified on the discrete restaurant dataset — a consistent 9-rule list with 0 training error (Patrons=None→No, Patrons=Some→Yes, …). Works on datasets with discrete attribute values. Closing as implemented.