Gearbox Failure Detector
Machine Learning classification project to detect gearbox failures. Involved data loading, EDA, feature selection, and model optimization to identify the best performing model.
Project Overview
The Gearbox Failure Detector is a predictive maintenance tool powered by machine learning. utilizing sensor data, it identifies potential faults in gearbox systems before they lead to catastrophic failure. This project demonstrates the application of data science in industrial settings to optimize operational efficiency and safety.
Key Features
Challenges & Learnings
"The dataset was highly imbalanced, with far fewer instances of failure than normal operation. This required techniques like SMOTE (Synthetic Minority Over-sampling Technique) and careful selection of evaluation metrics (F1-score instead of simple accuracy) to validly train the model."