Sales Model Of Travel Products And Its Application Based On Big Data

With the development of the Internet,a series of internet-based sales platforms have been emerged.The pattern of online sales makes transactions faster,provides more choices for users,provides a platform for merchants to sell,and changes the way we get data.We are now getting faster,more orders of magnitude,lower value density,and more complex forms of data than ever before.So we need to be able to handle big data and extract valuable information from it.Based on this background,this paper forecasts future product sales and analyzes the main factors influencing product sales through the processing and mining of the inherent attributes of products and users' purchasing behavior data.This paper used a travel website from January 2014 to November 2015 travel product information and product sales and prices of the real data as the training data,forecast in December 2015 to January 2017 various travel products in sales.First of all,the lack of data is observed and the appropriate method is selected according to the distribution of data to interpolate the missing value,and the variables with a large degree of loss are eliminated.The characteristics of the project,the next build model to deal with the data set of variables,select important variables and conform to the business scenario into the feature set,this article selects the following seven characteristics: the product of geographical features,date,rating and comment on features,price,features,holidays,month,order attributes.Then,the feature set is substituted into XGBoost model and GBDT model,and the sales of travel products are predicted,and the measurement of important variables in the model construction process is obtained.The XGBoost model and GBDT model are combined with linear weighting method,and the product sales are predicted again.XGBoost model predicted values and the real value of the mean square error(mse)between the lower and better prediction effect,and because XGBoost model of parallel computing capacity,greatly improving the model prediction speed,especially in the case of a large amount of data.Finally,based on the model prediction effect and variable importance measures,combined with actual situation on how to improve the travel product sales stimulus has consumption on the users' comments and suggestions of improving the user product grade.