search

firefly-cpp / NiaARM

A minimalistic framework for Numerical Association Rule Mining
MIT License
15 stars 6 forks source link

Add data squashing #85

Closed zStupan closed 1 year ago

zStupan commented 1 year ago
firefly-cpp commented 1 year ago

Are results similar to the results obtained by the conference paper?

zStupan commented 1 year ago

@firefly-cpp I tested with Abalone, and the results for cosine similarity are exactly the same. There's a slight difference with euclidean distance because of how the distance between categorical attributes is measured. Here I've just done one hot encoding, so the distance between different categories is 1. In uARMSolver it's 1/num_categories.

firefly-cpp commented 1 year ago

Perfect!