monitorai

The Official Python library for using Monitoring Machines Lab API

MonitorAI SDK

The Official Python SDK for Monitoring Machines Lab. MonitorAI is an observabiltiy framework designed for tabular ML to GenAI offering root cause analysis and decision intelligence allowing a linkup between ML and decision making for organizations that rely and use ML for solutions.

Getting started with monitorai sdk for feature drift :

• Install the package:

pip install monitorai

import pandas as pd
from monitorai import MonitorAI
from monitorai.drift.feature import CategoricalFeatureDrift

client=MonitorAI(api_key="***")

ref_df=pd.DataFrame({
"color":["red","red","blue","blue","red","yellow","yellow","red","blue"],
"social":["twitter","twitter","facebook","facebook","twitter","insta","insta","insta","insta"]})
    
analysis_df=pd.DataFrame({
"color":["red","red","red","blue","blue","yellow"],
"social":["twitter","twitter","facebook","insta","insta","insta"]
})


## create feature drift variable:
feature_drift=CategoricalFeatureDrift(ref_data=ref_df,analysis_data=analysis_df)

## get the bhattacharrya coefficients
bhatta_coeff_social=feature_drift.bhatta_coeff(feature="social")
bhatta_coeff_color=feature_drift.bhatta_coeff(feature="color")

## cauchy schwartz distance:
cauchy_schwartz_social=feature_drift.cauchy_schwartz_dist(feature="social")
cauchy_schwartz_color=feature_drift.cauchy_schwartz_dist(feature="color")

## euclidean distance:
euclid_social=feature_drift.euclidean_dist(feature="social")
euclid_color=feature_drift.euclidean_dist(feature="color")

## hellinger distance:
hellinger_social=feature_drift.hellinger_dist(feature="social")
hellinger_color=feature_drift.hellinger_dist(feature="color")

## jensen-shannon divergence:
js_social=feature_drift.js_divergence(feature="social")
js_color=feature_drift.js_divergence(feature="social")

## kl-divergence:
kl_social=feature_drift.kl_divergence(feature="social")
kl_color=feature_drift.kl_divergence(feature="color")

##psi:
psi_social=feature_drift.psi(feature="social")
psi_color=feature_drift.psi(feature="color")

Citing monitorai

If you use monitorai in your research, please cite it:

@software{monitorai,
  title = {monitorai: A Root Cause and Decision Intelligence Kit For Machine Learning},
  author = {Monitoring Machines Lab},
  year = {2026},
  url = {https://github.com/MasemeneMatlakanaBenny/monitorai},
  doi = {10.5281/zenodo.19646175},
  license = {MIT},
}