Mixed/Heterogeneous Data Support for constraint-based MB Algorithms

kuiy / CausalFS

An open-source package of causal feature selection and causal (Bayesian network) structure learning (C/C++ version)

58 stars 9 forks source link

Mixed/Heterogeneous Data Support for constraint-based MB Algorithms #2

Open arthii17 opened 3 years ago

arthii17 commented 3 years ago

I could find Markov Blanket algorithms supported only for either categorical or numerical data but not both (since only G2 & Fischer's test mentioned). Could you please clarify whether I missed to notice it or in future enhancement would be done in order to support the functionality?

cruyffturn commented 1 year ago

I'm also wondering whether hybrid graphs are supported. Supplementary results of "Causality-based Feature Selection: Methods and Evaluations" contains mixed datasets such as "madelon" with continues features and discrete labels.