/FWL-Metaheuristic

Feature Weights Learning metaheuristic for optimizing 1-NN classifier

Primary LanguageJupyter NotebookMIT LicenseMIT

Validate solutions

https://mh2223.danimolina.net/testsol.html

TODO

Task status
Organize datasets in a more ergonomic way (classes or some sort) OK
Implement 1-NN Classifier OK
Measure performance of classifier 1NN OK
Implement Greedy algorithm OK for now
Implement validation OK for now
Implement LS algorithm Needs Test...
Validity check tests Not implemented

IDEAS

  • fwl.py:78 Maybe do friends and enemies precalculation a bit faster

Notes

  • There's a possibility that a neighbour is exactly the same as the example being evaluated, in that case we will ignore this.

References