mlp9/Rossmann-stores-UK-sales-prediction-

Our project scope is to apply machine learning techniques to a real-world problem of predicting store sales. Germany’s largest store chain, has provided past sales information of 1115 stores located across Germany. We pre-processed, feature engineered the data, and examined 2 different machine learning algorithm for forecasting sales of store: Random Forest regression, and XGBoost. Then, we compared the method’s predictive power by computing Root Mean Square Percentage Error (RMSPE). We found that XGBoost model performed the best with a RMSPE score of 0.11 validation data set. Deployment has been done with Flask Web-App.

Jupyter Notebook