Overview

LC Prediction is a machine-learning project that turns raw driving signals into a simple answer at any instant: is the car keeping lane, changing left, or changing right?

The goal here is not just the prediction itself, but to design a general pipeline for:

• Cleaning and validating raw driving data
• Structuring it into supervised learning samples (sliding windows)
• Handling imbalance and preparing fair train/test splits
• Training and comparing models under consistent settings
• Evaluating results with metrics suited for rare events

This same structure can later be reused for future ML ADAS projects (e.g., predicting different maneuvers, labeling data, or testing alternative algorithms).

For details, continue to the Pipeline section.