This repository contains code for running various machine learning experiments on a dataset, with a focus on fault classification. The pipeline performs the following tasks:

• Preprocessing: Data scaling and feature selection
• Model Training: Several machine learning models including Random Forest, SVM, Naive Bayes, CNN, and ensemble methods (Stacking, Voting, and Bagging classifiers)
• Evaluation: Performance evaluation using accuracy score
• Visualization: Results visualization and model saving
• Logging: Logging the training process, results, and models

• Fault Type Mapping: Maps numerical fault types to descriptive labels for easier understanding.
• Custom Feature Selection: Feature selection using Mutual Information (MI) and Recursive Feature Elimination (RFE).
• Multiple Models: The pipeline supports multiple models like Random Forest, SVM, Naive Bayes, and CNN.
• Ensemble Learning: Includes methods like Stacking, Voting, and Bagging classifiers.
• Logging & Saving: Logs results and training logs to CSV and saves models.

The following Python libraries are required to run the code:

• pandas
• numpy
• scikit-learn
• tensorflow
• matplotlib
• os
• datetime

You can install the necessary dependencies with:

pip install -r requirements.txt