xgbGAMView utilizes the Generalized Additive Model (GAM) technique to establish an interpretable model and offers an easy-to-understand visual presentation of the model's predictive mechanism. The GAM technique combines predictors linearly, while allowing each predictor to have a non-linear relationship with the response variable. This flexibility enables the exploration of complex relationships within datasets, revealing trends and providing local explanations for predictions. class xgbGAMView employs decision tree algorithm implemented through "XGBoost", an optimized distributed gradient boosting library, and is designed for flexible modeling of regression, binary classification, and survival analysis tasks. The visual presentation, represented by figures of the features' shape functions, allows for easy examination and comparison of how different features influence the predicted values.

TREE-Gs are decision trees specialized for graph data. They can be used for classificaiton, regression, vertex-labeling, graph-labeling, and edge-labeling. The model is described in the paper TREE-G: Decision Trees Contesting Graph Neural Networks. TREE-G is highly recommended when learning over tabular features, and often outperform Graph Neural Networks on such tasks, as shown in the paper. The library is Scikit compatible.

Unit-test for data to support data centric AI

Auto-assessment tool for teaching and learning computational thinking. It is a python library that allows teachers to create assignments that allow students to explore multiple solutions while getting feedback on their work.

Measuring Feature Importance of data to allow explainability. This tool uses the MCI approach to measure feature importance for the global setting when a scientists are trying to understand the contribution of different features in the data to a given outcome.

This tool allows compressing almost any machine learning and deep learning model to any type of small model

In many applications of machine learning records of data contains sets of items. For example, medical records may contain sets of medications and astrological data may contain sets of stars. Set-Tree is a python tool that allows applying decision trees and boosted decision trees to such data.

CryptoNets allow applying Neural-Networks to data that is encrypted with homomorphic encryption to preserve privacy. The data remains encrypted during the entire inference process

A margin based feature selection algorithm implemented as a matlab library