search

ModelOriented / localModel

LIME-like explanations with interpretable features based on Ceteris Paribus curves. Now on CRAN.
https://modeloriented.github.io/localModel
14 stars 3 forks source link

Change discretization method #30

Closed mstaniak closed 3 years ago

mstaniak commented 5 years ago

Decision trees suck

mstaniak commented 5 years ago

This is what the quality of interpretable features depends on.

mstaniak commented 5 years ago

Changepoint doesn't seem much better than decision trees

mstaniak commented 5 years ago

A comparison of discretization methods would be nice, including correlation-aware discretization and interactions-aware discretization