Summary
- Try the live demo here
- It is hosted on Dash Enterprise
- It uses IBM CODAIT’s open-source AIX360 library
- The code is open-sourced on Github with instructions on how to run it locally
- Interested in building AI apps like this? Reach out!
Description
This app shows you how to visualize and analyze explainable AI (xAI) models with Dash and IBM’s AI Explainability 360 (AIX360). It uses a LogisticRuleRegression model trained on the UCI Heart Disease data set for predicting the presence of heart disease in Cleveland patients.
The model we showcase is not only accurate, but it also lets you compute the coefficient of each rule, which is the combination of various conditions of the patient. Furthermore, you can also visualize the generalized additive model component of each patient attribute. According to the docs, the latter " includes first-degree rules and linear functions of unbinarized ordinal features but excludes higher-degree rules."
Dataset
You can find information about each of the input feature from the original data repository.
Once thing that was not clear was the definition of S-T Segments. To learn more about that, you can check out this article.