Weekly Sales Prediction of Walmart Stores Based on Three Regression Models

Abstract

In the information era, the sales of a supermarket are influenced by a lot of factors. It is very important for a supermarket to predict its expected sales according to several given data about the factors and then make corresponding plans. Regression method is a common method to do the prediction and it applies the known data to set up regression models to predict the future data according to some given information. This study will try to find out the relationship between the factors and the weekly sales of two Walmart stores and try to use the model to predict its future sales by using three regression models, the linear regression model, the elastic net model and the SVR model. Then, the paper will divide the raw data into a training set and a testing set and build the model with the data in the training set and find out the best one by calculating the MAPE of the testing set for every model. According to the analysis, three SVR models and one elastic net regression model perform best in four situations of the prediction and do a simple analysis to every factor about their negative or positive impact on the weekly sales and their weight. These results will offer a prediction about the future sales of the two Walmart stores for the market and can be used as a reference for both the management of the market and the consumers.