xai-analytics

Cross-field empirical trends analysis of XAI literature

Code for paper "Towards Training GNNs using Explanation Directed Message Passing"

XAI-Analytics is a tool that opens the black-box of machine learning. It helps the user to understand the decision-making process of machine learning models.

Replication package for the KNOSYS paper titled "An Objective Metric for Explainable AI: How and Why to Estimate the Degree of Explainability".

The notebook on the main topic of interpretable machine learning is a descriptive and instructive analysis of a car data set from a public source.

This repository aims to explain state-of-the-art Fake News Detection models. Huggingface module is intended for explaining only-text-based fake news detection models in Transformers. GNNFakeNews module is an attempt to explain GNNs that are hybrid models for fake news detection using GNNExplainer

RAD-XAI is an explainable-AI framework for chest X‑ray pneumonia detection that benchmarks CNNs and ViTs using Grad‑CAM-based saliency maps and quantitative XAI metrics (concentration, faithfulness, agreement) to audit model reasoning and surface clinically unsafe failure modes beyond standard AUC/accuracy.

Open source research documents, free to use or share or modify. Enjoy!

This repository focuses on classifying cancer types using ML algorithms. It includes data preprocessing, feature selection, model training, and evaluation using Logistic Regression, Decision Trees, Random Forest, SVC, KNN, and Naive Bayes. Explainable AI techniques like SHAP and LIME are implemented for transparency.

Anomaly Detection Web Application using Dynamic Ensembles and SHAP Explanations

Projects for Praxis Data Science

https://github.com/jphall663 -- https://github.com/jphall663/awesome-machine-learning-interpretability -- https://www.youtube.com/watch?v=d6j6bofhj2M&list=PLV8yxwGOxvvovp-j6ztxhF3QcKXT6vORU&index=3

Explainable AI Streamlit app for customer segmentation using a Gaussian Mixture Model (GMM) with SHAP-based explanations via an XGBoost surrogate. Includes interactive pages for cluster prediction, local and global explainability, probabilistic outlet-city prediction, and city-aware anomaly detection, with saved model artifacts.

PRISM, an explainable risk assessment tool. Contributing factors are estimated using a lightweight, real-time approximation method. PRISM is a research and demonstration tool and is NOT YET validated for clinical use or real patient decision-making.