Store-sales Forecasting Model to Determine Inventory Stock Levels using Machine Learning

Abstract

Predicting sales had been a common practice. Sales prediction plays a crucial role in the business world. It gives accurate and dependable information related to current and previous events, also the events that are expected to occur in the future. However, because traditional sales approach lack insight into customers' buying patterns, they can no more help businesses in keeping up with the pace of a competitive business world. Machine Learning evolvements have resulted in major changes in sales and marketing fields. Many key aspects like consumers' buying patterns, target audiences, and estimating sales for upcoming years can be determined easily, all thanks to the advancements in machine learning and thus helping the sales team in the companies for making plans for a boost in sales.In the proposed methodology, the study of several forecasting methods used in the forecasting of the future sales of stores keeping previous year's sales in view. We tried linear regression model, Random Forest and XGBoost regressor. But linear model performed poor, so it is not included here. The prediction models implemented herein are random forest regressor and XGBoost regressor. Both of the regression techniques give better accuracy and less RMSE (Root Mean Squared Error) value than linear model. XGBoost perform best in all the three models.