Regression on Bike Sharing Demand

• Implemented multiple linear and polynomial regression models (degrees 2-4) from scratch using the Normal Equation, without regression libraries, to predict hourly bike sharing demand.

• Performed feature engineering, normalisation, and an 80-20 train/test evaluation, comparing models using MSE and R2, and identified the optimal trade-off between accuracy and complexity.

• Delivered a research-style report and Kaggle-ready prediction output as part of an Optimisation project.

• Used Dataset: Bike Sharing Demand, Kaggle (2014)

• Language: Python

• Skills: Optimisation, Regression