1️⃣ Performed extensive data preprocessing, including One Hot Encoding and Label Encoding on a dataset of 547,517 rows and 100 columns, enhancing feature representation.
2️⃣ Implemented cutting-edge algorithms such as Linear Regression and RandomForestRegressor, achieving a remarkable 25.1% improvement in Root Mean Squared Error (RMSE) through log transformation of the dependent variable, demonstrating a keen understanding of predictive modeling.
3️⃣ Employed the powerful RandomForestRegressor ensemble algorithm to predict customer purchase amounts, achieving a notable 36.2% reduction in RMSE and a 16.1% boost in R-squared (R2) after log transformation.
4️⃣ Leveraged expertise in ensemble learning to outperform traditional linear models, showcasing a strategic approach to model selection for improved accuracy in predicting retail purchase behaviour.