Analysis And Forecast Of Factors Affecting Online Travel Product Sales

With the rapid development of economic society and information technology,the way in which information is disseminated has changed dramatically.As a new tourism business model,online tourism is becoming more and more popular with consumers.The product information of online travel websites is rich,including various information such as prices,user feedback,itinerary features,preferential activities and service guarantees.Each indicator affects the sales of online travel products more or less,and it is especially important to find out the factors that significantly affect sales and make effective forecasts.This article takes the online travel products on the Qunar website as the research object,and uses the octopus collector to collect the basic information of the online travel products on the Qunar website.After the data was pre-processed,a descriptive analysis and a single-factor impact analysis were first performed to visually show the impact of each factor on the sales of online travel products.Secondly,use the method of ordinal regression to transform the sales volume into an orderly factor,that is,high sales volume,medium sales volume,and low sales volume,and identify 11 key characteristics that affect online travel product sales,including user satisfaction,travel agency credit rating,destination,travel days,price,type of travel,travel agency popularity,type of transportation,whether it includes team discounts,membership prices,and no self-pay.After that,the results of the model are explained from price,user feedback,itinerary features,preferential information and service guarantee information.Then,establish a multivariate linear regression model that can explain the influence on the sales of online tourism products.All the tests of the model have passed.The adjusted R2 is 0.7735,and the degree of fitting is good.Further,according to the regression coefficients,each influencing factor is quantitatively described.It is believed that factors such as user satisfaction,whether it includes team discounts and membership prices have the most significant impact on the sales of online travel products.Finally,the sales volume of online travel products is predicted.In order to improve the accuracy,a gradient lifting regression tree model(GBRT)and a random forest model(random forest)are established and compared according to the key factors selected by sequential regression.After parameter tuning,the RMSE of GBRT model is 0.0542,the deviation is small,and the mean square error is greatly reduced compared with multiple regression model,from 0.2988 to 0.0341.After calculation,the accuracy of the GBRT model is as high as 98.80%.The mean square error of the random forest model is 0.400,which is higher than the GBRT model;the accuracy rate is 97.24%,which is lower than the GBRT model.Therefore,the GBRT model can more accurately predict the sales volume of online travel products in this article.This paper studies the factors that affect the sales of online travel products by establishing statistical models and algorithmic models,and makes more accurate predictions of sales based on these influencing factors.The aim is to summarize a method for consumers to evaluate the sales volume of online travel products,help them make more appropriate decisions,and also help to improve the online travel product evaluation and sales system in China.At the same time,it provides constructive suggestions for online travel websites and travel agencies to continuously improve the online travel market.