Research On The Order Forecast Of A Restaurant O2O Enterprise Based On The ARIMA-BP Combination Model

With people's constant pursuit of quality of life and the advent of the Internet+ era,many companies emerged that rely on the 020 business model to provide life services,such as Xinmei company,Qunar,58 Home,and Chain Home.These companies have made the Internet a frontline for offline transactions through a combination of online and offline services,making it easier for users to meet their needs,such as clothing,food,housing,and transportation.Among them,catering is an important vertical classification in the 020 field.The main display forms include group purchase and takeaway.On the one hand,the market competition in this field is extremely fierce,and the external environment changes rapidly.On the other hand,the order volume of enterprises directly affects the sales and market share of the company.Therefore,for the enterprise,it is necessary to continuously improve the ability of data-based operations to respond to market changes.How to predict orders more quickly and accurately,predict the future development environment,and tap potential market demands will be very important.The traditional demand forecasting model is generally based on the time series algorithm model.It uses the relationship between historical sales and time for modeling and forecasting.It is generally aimed at the manufacturing and retail industries that sell actual goods as the carrier,but less research to the 020 catering company's orders prediction.For catering 020 companies,the biggest feature is the online platform,with a number of offline catering companies to cooperate,as many as possible to enter more businesses on the online platform.On the other hand,in business promotion and marketing methods,it is totally different from traditional offline catering companies,giving users the means to set marketing strategies and strategies for catering is the survival magic of 020 companies.The research on order forecasting for catering 020 companies can quickly help companies understand the future order trend and assist in the formulation of marketing strategies.This paper takes a catering 020 company in China as background,and establishes the forecasting model with the goal of predicting enterprise service orders.This paper analyzes the existing business model and business data of catering 020 companies,and takes the specific business of an 020 catering company as the background.It finds that the order data has a certain trend and seasonal characteristics,and has a part of linear characteristics;in addition,the change of order quantity is also related to factors such as promotional promotions that businesses participate in,business experience,and so on.It is also affected by nonlinear factors such as the external environment,weather,and competitors and so on.Therefore,this paper selects a more representative ARIMA time series and BP neural network model,and establishes a combined forecasting model.This model is based on three different prediction scenarios,it was implemented and verified.The combined model can not only satisfy the prediction of linear feature data,but also satisfy the prediction of nonlinear feature data.Finally,the article compares and analyzes the prediction results of the model.The experiment finds that the establishment of the combined forecasting model has achieved a better forecasting result and has achieved the expectation of improving the forecasting accuracy of the order.The research of this article can help the enterprises under the O2O mode of catering to in-depth data operation,and understand the trend of future orders in different dimensions in advance,which can assist the employees of all levels in the company to carry out reasonable resource allocation,strategy decentralization and future planning in the business development process.