PHIK

Phik is an open-source Python library for measuring statistical association between categorical and discrete variables. It centers on the phik statistic, a generalization of classical measures such as the chi-squared test, Cramér’s V, and mutual information. Phik is designed to capture a wide range of dependencies, including nonlinear relationships that traditional correlation measures may miss.

The library provides a convenient way to compute pairwise associations across multiple variables. It can generate a phik matrix for a dataset, where each entry represents the strength of association between a pair of variables. It is built to work with pandas DataFrames, supports categorical and discrete data, and can handle missing values. Users can apply it to large datasets and obtain a compact summary of inter-variable relationships suitable for exploratory data analysis, feature selection, or network visualization.

Usage is straightforward: phik is installed from the Python Package Index and then imported in Python code. A typical workflow involves loading data into a DataFrame and calling a function such as phik_matrix (or equivalent API) to produce the pairwise phik scores. The results can be used to rank variable pairs by association strength, identify clusters of related features, or guide subsequent modeling decisions.

Phik is commonly used in data science and analytics workflows to assess relationships in categorical data, especially when traditional correlation metrics are inadequate. It complements other information-theoretic and statistical measures and is part of open-source ecosystems for data exploration and feature engineering. See also: Cramér’s V, mutual information, chi-square test.