Contextualized Logo

Contextualized.ML

A statistical machine learning toolbox for estimating models, distributions, and functions with context-specific parameters.

State-of-the-art methods to estimate and explore context-specific models

Comprehensive support for multiple types of models and meta-models, with training and analysis tools

Building blocks to support extending meta-models to contextualize your application-specific mdoel

Why ContextualizedML?

Easy usage

Access state-of-the-art techniques through an open unified API and rich visualizations.

Flexible and customizable

Understand models using a wide range of explainers and techniques using interactive visuals. PyTorch Choose your algorithm and easily experiment with combinations of algorithms.

Community-driven open source toolkit

Join the community, request and contribute improvements to make your model the best it can be.

Model Types

Contextualized Regression

Do the features which predict an outcome change from sample to sample?

Contextualized Bayesian Networks

Does the joint distribution of features change from sample to sample?

Contextualized Correlation

Do the correlations between features change from sample to sample?

What You Can Do With ContextualizedML

Find Hidden Heterogeneity

  • Identify changes in data distributions with Contextualized Bayesian Networks
  • Identify changes in feature correlations with Contextualized Correlation Networks
  • Connect Heterogeneity to Outcomes

  • Identify Heterogeneous predictors of outcomes with Contextualized Regression
  • Cluster samples according to reasons for outcome
  • Who Can Benefit from ContextualizedML?

    Data Scientists

    Understand heterogeneity hiding in your dataset.

    Scientists

    Understand heterogenetiy in the observed populations.

    ML Engineers

    Build models which adapt to new environments and contexts.

    ML Researchers

    Contextualize your new models.

    Getting Started

    Install ContextualizedML

    Contribute to ContextualizedML

    We encourage you to join the effort and contribute feedback, algorithms, ideas and more, so we can improve the toolkit together!

    Contribute