Research On Influencing Factors And Prediction Method Of Takeout Order Based On A Subway APP

In the new era,as a new media,rail transit app has strong communication and promotion ability.At the same time,as a service-oriented APP with a large number of users and high dependence,rail transit APP has also become a diversion entrance for some enterprises.With the increasingly fierce competition of enterprise products,enterprises have used new media to broaden marketing channels.It is very important to improve and innovate the advertising communication mechanism in new media,constantly update the products iteratively,grasp the psychological needs of consumers,and let the products better convey the product advantages to consumers with the help of new media.Therefore,taking the takeout business of a subway APP as the background,the thesis analyzes the takeout order data of the takeout enterprise in the APP,which is of great significance to improve the income of the enterprise and improve the marketing mode.The thesis mainly studies the influencing factors of takeout orders on a subway APP platform,as well as the interaction between takeout orders and subway APP advertising clicks,and analyzes and forecasts the takeout orders of takeout enterprises in the subway APP.The specific process is as follows: firstly,the actual order data is analyzed based on the development status of takeout enterprises in rail transit APP and relevant literature research,and the main factors affecting the takeout order volume of subway APP are selected.Then the Granger causality test method is used to study the causality of the main influencing factors.Then,from the perspective of discussing the interaction between subway APP and takeout enterprises,the impact mechanism between subway APP takeout advertising click and takeout enterprise order is studied by using impulse response function and variance decomposition method.Finally,ARIMA model,VAR model with explanatory variables and multi feature LSTM neural network model are established to predict the takeout order volume of subway APP.According to the linear and nonlinear characteristics of the original order volume series,the time series model with better prediction effect is selected for combined prediction with LSTM model,and the accuracy of different models is compared.The results show that among the main influencing factors,there is a bi-directiona causal relationship between the clicks of subway APP takeout advertising and the number of takeout orders.Both sides have a positive impact on each other.The contribution rate of takeout advertising clicks to the change of takeout orders is 16.12%.The contribution rate of takeout orders to the change of takeout advertising clicks on the platform is 31.02%,indicating that the advertising clicks on the APP platform are greatly affected by the orders of takeout enterprise.In addition,the comparative analysis of the prediction model shows that the prediction effect of VAR model with explanatory variables is better than that of univariate ARIMA model,so the VARLSTM combination model is established.The VAR-LSTM combined model has better prediction effect than other single models,which is of guiding significance to the takeaway enterprises’ marketing operation management in the subway APP.