scientistmetrics : python library for model metrics

About scientistmetrics

scientistmetrics is a Python package for metrics and scoring : quantifying the quality of predictions

Why scientistmetrics?

Measure of association with categoricals variables

scientistmetrics provides the option for computing one of six measures of association between two nominal variables from the data given in a 2d contingency table:

• Chi - squard test : https://en.wikipedia.org/wiki/Chi-squared_test
• Cramer's V : https://en.wikipedia.org/wiki/Cram%C3%A9r%27s_V
• Tschuprow's T : https://en.wikipedia.org/wiki/Tschuprow%27s_T
• G-test : https://en.wikipedia.org/wiki/G-test
• Phi coefficient : https://en.wikipedia.org/wiki/Phi_coefficient
• Pearson contingence coefficient : https://www.statisticshowto.com/contingency-coefficient/

Classification metrics

scientistmetrics provides metrics for classification problem :

• accuracy score
• f1 score
• precision
• recall
• etc...

Regression metrics

scientistmetrics provides metrics for regression problem :

• Rsquared
• Adjusted Rsquared
• Mean squared error
• etc...

Powerset model

scientistmetrics provides a function that gives a set of all subsets model.

Notebook is availabled.

Installation

Dependencies

scientistmetrics requires :

python >=3.10
numpy >=1.26.4
pandas >=2.2.2
scikit-learn >=1.2.2
plotnine >=0.10.1
statsmodels >=0.14.0
scipy >=1.10.1

User installation

You can install scientistmetrics using pip :

pip install scientistmetrics

Author(s)

Duvérier DJIFACK ZEBAZE duverierdjifack@gmail.com