Research of Commodity Sales Prediction Based on GWO-Prophet
The business activities of retail enterprises are inseparable from the sales of goods.The sales forecast of goods pro-vides an important basis for enterprises to formulate production plans and business decisions.Aiming at the problem that the time se-ries of sales volume in enterprise sales forecast is greatly affected by external conditions and the prediction accuracy is low,this pa-per proposes a commodity sales forecasting method based on GWO-Prophet.Based on the sales data of a retail enterprise from 2015 to 2018,the Prophet model is used to construct the low-dimensional time series feature components of the corresponding trend item,seasonal item,holiday item and residual item through the high-dimensional sales data of the Prophet model.After fitting these low-dimensional feature components,the sales data of the next year are predicted by the addition model.Through the grey wolf optimization algorithm(GWO),the Prophet model parameters are intelligently optimized to prevent the model from falling into local optimum and improve the accuracy of the model.The Prophet model optimized by the grey wolf optimization algorithm can bet-ter fit the influence of external factors such as mutation points,seasonal items,holiday items on sales.MAE,MAPE and RMSE are used as indicators for model evaluation.The results show that the prediction accuracy based on GWO-Prophet model is not only bet-ter than the single Prophet model,but also better than other comparison models such as ARIMA,SARIMA and LSTM.
Prophet modelGWO algorithmtime seriessales forecastdecomposable model
国家科技期刊平台
NSTL
万方数据
