ML Model Pipeline

This repository contains two core files:

• preprocess.py: Preprocesses your raw dataset (handles missing values, scaling, label management).
• xai.py: Trains machine learning models on preprocessed data, evaluates performance, and generates feature importance and explainability reports.

---

Requirements

pip install pandas scikit-learn matplotlib eli5 numpy

---

Step 1: Data Preprocessing

Ensure your raw data CSV (e.g., device1top20features.csv) is in the same directory. Edit preprocess.py to specify your label column if it is not named "label".

Run preprocessing: python preprocess.py

• Input: device1top20features.csv
• Output: preprocesseddevice1features.csv

---

Step 2: Model Training, Evaluation & Explainability

After preprocessing, run the modeling script: python xai.py

• Input: preprocesseddevice1features.csv (output of Step 1)
• Outputs:
  • Performance metrics (accuracy, precision) for Decision Tree, Random Forest, Neural Network (printed to console)
  • Feature importances plot: featureimportances.png
  • Model explainability reports: HTML files (eli5explanation<MODEL_NAME>.html) for each trained model, containing interpretable weights/rules and example prediction explanations

---

Notes

• Check the top of both scripts if you need to adjust filenames or parameters.
• The HTML explanation files are viewable in any browser for human-readable model insights.
• Make sure Python 3.x is installed.