Leveraging Machine Learning for Accurate Store Sales Prediction: A Comparative Study

Accurately forecasting store sales is essential for optimizing inventory management, improving customer satisfaction, and driving revenue growth. This paper explores the application of machine learning and deep learning models, including Linear Regression, AdaBoost, Random Forest, XGBoost, SARIMA, ARIMA, and Prophet, to predict future sales based on historical data. By leveraging two distinct datasets-Superstore Sales and Walmart Store Sales-the study aims to provide a comprehensive evaluation of model performance using various metrics such as R2, MAE, RMSE, and MAPE. Our findings indicate that Linear Regression with AdaBoost performs best on the Superstore dataset, achieving an R2 score of 77%, while ARIMA outperforms other models with an R2 score of 93% in the forecasting category. On the Walmart dataset, Random Forest and Linear Regression tied, achieving the highest accuracy with R2 scores of 97%. These results demonstrate the effectiveness of machine learning models in sales prediction and offer valuable insights for retail decision-making.