Sales Forecasting for Rossmann Store using GRU with Weather Features

M VIJAYA KUMAR , LANKE GAYATHRI , DEVARAMPATI SNITHA GRACE , SHAIK ARSHAD , VENKATA GANESH KUMAR DEGALA

Sales Forecasting, Rossmann Store, Gated Recurrent Unit, Deep Learning, Weather Features, Grid Search, RMSE, Time Series Prediction, Retail Analytics, Demand Forecasting.

Sales forecasting plays a vital role in ensuring the success and efficiency of retail businesses. This project focuses on predicting the daily sales of Rossmann stores using a Gated Recurrent Unit (GRU)-based deep learning model. The dataset used comprises multiple years of historical sales data along with store information, promotions, holidays, and weather features. The inclusion of weather features serves to improve the forecasting accuracy, as external factors like temperature, precipitation, and wind speed can influence customer footfall and purchasing behavior. The GRU model was chosen for its ability to handle sequential data effectively and capture long-term dependencies. Grid Search was employed to fine-tune the hyperparameters and enhance the performance of the model. The model was trained and evaluated using Root Mean Square Error (RMSE) as the performance metric. Results demonstrate that incorporating weather features into the GRU-based model significantly improves forecasting performance, making it a practical solution for demand prediction in the retail sector.

Volume 10 Issue 9, September-2025

Pages : a103-a107