Research And Application Of Prediction Technology For Men's Wear Sales Data

With the development of the apparel industry,the Chinese men's wear industry has become an important part of the apparel industry and a part of the competition.In order to be in a favorable position in the competition,it is necessary to understand the market demand and seize the opportunity,which requires an accurate forecast of the future sales of men's clothing.The traditional clothing sales forecast based on empirical judgment and market research analysis lacks scientific basis and the factors considered are relatively simple,which cannot make the prediction result accurate.The paper studies the clothing sales forecasting technology from two aspects: single forecasting model and combined forecasting model.Based on the analysis of the characteristics of clothing sales data,this paper studies the sales data forecast based on a single model on the basis of reviewing the relevant theory and technology of data prediction.According to the linear characteristics of clothing sales data,the paper uses ARIMA model to predict men's sales data.For the nonlinear characteristics of clothing sales data,the paper uses neural network model to predict men's sales data.There are trending linear characteristics of clothing sales data,and there are also nonlinear characteristics due to factors including brand,popular factors,price,and season.The single prediction model has certain limitations.The paper further studies the related prediction techniques based on the weighted combination prediction model.Through the experimental analysis,the method of calculating the weight of each part of the combined forecasting model by using the optimal weight method is determined.Using the actual data,the three kinds of combinations of different models are predicted and analyzed.The results show that the ARIMA-BP-RBF combined model has the best prediction accuracy.Specifically,the paper includes the following work.(1)The research status of clothing sales data forecasting is reviewed.Through the analysis of the actual sales data of a certain merchant,the linear and nonlinear characteristics of men's clothing sales data are summarized.Through data aggregation,validity screening,missing values and outlier processing,high-quality data was obtained,which prepared for the data prediction work.(2)The clothing sales data forecast based on the ARIMA single model was studied.Combined with the characteristics of men's sales data,the ARIMA modeling process was analyzed.The non-stationary data was converted to a stationary state by differential processing,and the model parameters were determined by pattern recognition and verification.Experiments show that the ARIMA model predicts the linear part of the data accurately,and there are still some shortcomings in the overall menswear sales data forecast.(3)The prediction of clothing sales data based on a single model of neural network is studied.Combined with the nonlinear characteristics of menswear sales data,the BP neural network and RBF neural network modeling process are analyzed.By analyzing the characteristics of various transfer functions,it is determined that the Sigmoid function is suitable for the BP neural network model,and the Gaussian function is applicable to the RBF neural network model.Through the analysis of the principle of neural network model,the number of nodes in each layer of the neural network is determined,and the sample data is trained and tested.Experiments show that the BP neural network model and the RBF neural network model predict the nonlinear part of the data accurately.Because of the linear part of the data,there is still a shortage of overall menswear sales data prediction.(4)The clothing sales data prediction technology of ARIMA and neural network combined model was studied.Through the experimental analysis of multiple weighted combination methods,the method of using the optimal weight method as the weight of each part of the combined model is determined.The ARIMA-BP combination model,the ARIAM-RBF combination model,and the ARIMA-BP-RBF combination model were constructed by the optimal weight method.The experimental data analysis was carried out to determine that the ARIMA-BP-RBF combination model is most suitable for menswear sales forecast.