Walmart Sale Forecasting Model Based On LSTM And LightGBM

Abstract

Accurate prediction of sales can help companies adjust production strategies in a timely manner, improve production efficiency, and improve industry competitiveness. This article is based on LSTM and LightGBM models to realize Wal-Mart's sales forecast for a period of time. Considering that the amount of data given by the materials is huge and inconsistent, we have also carried out feature engineering processing on the data, unifying abnormal data and extracting data features, so as to obtain processed data that is convenient for modeling, and then use the LSTM model to Learn and forecast sales. Experiments show that the RMSSE of the proposed model in this paper is only 0.821, which has better predictive ability than traditional linear regression models and SVM models.