nicodv / kmodes

Python implementations of the k-modes and k-prototypes clustering algorithms, for clustering categorical data
MIT License
1.23k stars 416 forks source link

add feature weights (w-k prototypes) #200

Open joaquin-tempelsman opened 4 weeks ago

joaquin-tempelsman commented 4 weeks ago

Based on proposed approach by Çörekci, Osman, and Ayla Saylı. "A New Clustering Algorithm of Hybrid Data According to Weights of Attributes." Avrupa Bilim ve Teknoloji Dergisi 5.9 (2016): 28-37.

Implemented feature adds feature weights by passing a weight argument per categorical and numerical feature. This weight is used by both default similarity functions (euclidian for numerical and matching for categorical). Assertions checking for correctly passed weights are added under init and fit methods.

Pending: include integration tests