Conditional Independence Tests#
FastCIT has twelve conditional independence tests. All tests share one API (see one API for all tests). A test checks whether X and Y are independent given a set Z of other variables. HSIC has no Z: it checks only whether X and Y are independent. Select a test to see an example in each language and the method. Each language makes its own random data, so the values differ between the language tabs. With the same data, all languages give the same result.
Test |
Data type |
|---|---|
Chi-squared ( |
Discrete |
Log-likelihood ( |
Discrete |
Cressie-Read ( |
Discrete |
Freeman-Tukey ( |
Discrete |
Modified log-likelihood ( |
Discrete |
Pearson correlation ( |
Continuous |
Fisher-z ( |
Continuous |
Pearson equivalence ( |
Continuous |
HSIC ( |
Continuous |
KCI ( |
Continuous |
GCM ( |
Continuous |
Pillai’s trace ( |
Mixed (discrete and continuous) |