XAI Competition 2022

Inspiration

I was inspired to work on this project through the XAI competition organized by the ALPS lab and AI Society at UT Dallas.

What it does

For this project, I applied the Fold-R++ (for binary classification) and Fold-RM (for multiple-category classification) to various Kaggle datasets to test the algorithms and analyze the outputs.

Datasets used:

1. https://www.kaggle.com/laavanya/stress-level-detection
2. https://www.kaggle.com/yasserh/breast-cancer-dataset?select=breast-cancer.csv
3. https://www.kaggle.com/adityakadiwal/water-potability
4. https://www.kaggle.com/tejashvi14/employee-future-prediction

Learn more about the tools used:
Fold-R++: https://github.com/hwd404/FOLD-R-PP
Fold-RM: https://github.com/hwd404/FOLD-RM

Challenges I ran into

1. Understanding how to install and use the tools.
2. Analyzing the rules outputted by the algorithms.
3. Finding bias/unfairness in the algorithms.

Analysis

It was interesting to see the difference in the rules that were generated by the algorithms. For the first dataset, the rules were very simple and the accuracy is pretty high whereas the rules for the other three datasets were a lot more complicated with a lower accuracy. With the employee dataset, there were a lot of exceptions and only 2 rules.

Built With

• fold-r++
• fold-rm
• python