Research On Dynamic Prediction Method Of Life Cycle Curve Of Fashion Products

With the rapid development of the manufacturing industry,the changes in people’s consumption levels and consumption habits,the life cycle of more and more commodities is becoming shorter,especially for some fast fashion brands.In order to better meet the consumer’s consumer demand and to bring greater benefits to the company,it is very necessary to predict the sales of short life cycle fashion products.With the rapid development of information technology,the processing power of big data has also increased.Data analysis and mining is to explore the correlation between data on the basis of a series of related disciplines such as statistics,database,neural network and artificial intelligence.Used to assist decision making.Based on the current background,we use data analysis and mining methods to predict the sales of short-life fashion products.Many scholars have done a lot of research work on the prediction of short life cycle fashion products.Based on the product life cycle curve,this paper predicts the entire life cycle curve of the product.Based on the research foundation and method of the predecessors,this paper proposes a prediction method using time series clustering,which links the time series with the product life cycle curve to predict the sales volume of the product throughout its life cycle.Propose your own solutions to the problems in the short-life cycle fashion product sales forecast.In view of the lack of volatility in sales time series data,we first perform curve fitting on time series.The curve model uses Bass model,polynomial model,piecewise function model,etc.,by minimizing the root mean square error of predicted value and real sales.(RMSE),get the parameters of each curve.Select the best fit curve based on RMSE,log likelihood and Akaike information criteria(AIC)and generate a new time series;in addition,in order to get better prediction results,based on product sales time Sequences are clustered,and clustering clustering is used to obtain better clustering results.For the classification of new products,the traditional classification method is to divide the new products into certain categories,because the difference between similar fashion products is not very large,so this paper uses fuzzy classification method to calculate new products for each category.The degree of membership is not clear about the category of new products,making full use of the product information of each category;the weekly sales forecast of new products is predicted by static and dynamic update methods,and the fitting sales curve of each category is based on the membership.The weighted average is used as the basis for the new product forecast,and the weekly sales forecast value is added to predict the total sales of the new product.The degree of membership of static prediction is the degree of similarity between the new product and the historical product attribute.The dynamic update forecast takes into account the sales information known to the new product,and considers the similarity between the sales volume and the historical product sales sequence,so that the future can be predicted more accurately.Sales.Finally,using the consumption data of a fashion enterprise to conduct empirical research,the results show that the dynamic forecasting effect of weekly sales is more effective than the static forecasting effect.The total sales forecast is after multiple linear regression,multiple nonlinear regression,random forest,time series clustering.After the comparison experiment of the multiple linear regression model,the results show that the method of using the life cycle curve to predict the total sales volume is significantly improved compared with the comparative experiment method,which also verifies the validity of the model idea.The sales forecast for new fashion products is of great significance in real production.The total sales forecast can estimate the total demand for new products listed,which guides the production of new products,guides production scheduling,and arranges appropriate production capacity.It is important to make decision-making significance;the practical significance of weekly sales forecast is to arrange inventory and distribution volume in each region.Reasonable inventory arrangement can not only reduce the inventory cost but also avoid the shortage of goods caused by unreasonable goods and the dissatisfaction of customers.Therefore,the text has important practical significance.