Research on sales Forecast based on XGBoost-LSTM algorithm Model

Abstract

Reasonable sales forecast is very important for enterprises. The short-term and long-term sales changes of a product are helpful for enterprises to make marketing strategies and sales decisions. On the basis of in-depth analysis of the characteristics of a certain algorithm model and long and short memory neural network, and according to the data set provided by a supermarket chain in kaggle competition, a XGBoost-LSTM neural network combination model for sales forecasting and a classical time series prediction model are constructed to compare the experimental results. The experimental results show that the XGBoost-LSTM neural network prediction model has higher accuracy than the time series prediction model, which can provide an important scientific basis for the supermarket chain to make sales forecast.