Design And Implementation Of O2O Coupon Usage Forecast System Based On LightGBM

The O2O model is an e-commerce model that combines offline transactions with the Internet.In the O2 O model,online malls provide online users with various types of information about offline stores by information providing services.After obtaining relevant information about the store,users can place orders and pay online,and then use the order information to pick up products from offline stores or enjoy the service.As a means of attracting consumers,coupons have always been the preferred solution for offline store marketing.But for most consumers,random coupons can cause meaningless or boring distractions.On the other hand,for offline stores,a large number of random coupons will not only reduce the credibility of their own brands,but will also make it difficult to control marketing costs.Compared with the traditional gradient boosting decision tree algorithm and XGBoost,the LightGBM algorithm not only has faster training efficiency,lower memory usage,and higher accuracy,but also supports parallel learning and large-scale data processing.At present,there are few researches on the application of LightGBM algorithm in related fields.This article uses the consumer offline transaction information and online click information provided by the Ali's Tianchi contest platform to establish an O2 O coupon usage prediction model based on the LightGBM algorithm.The validity of the model is verified through experiments.Based on this model,a O2 O coupon usage prediction system is designed and implemented based on Spark,which is a big data computing framework.The works of this article are as follows:First,this article achieve the personalized delivery of coupons by establishing a prediction model of O2 O coupon usage.In addition to giving merchants stronger marketing capabilities,the personalized placement of coupons can also give consumers with certain preferences real benefits.Second,there is not much research literature on the LightGBM algorithm at home and abroad.This article uses the LightGBM algorithm to predict the use of O2 O coupons and expand the application field of the LightGBM algorithm.Third,this paper developed an O2O coupon usage prediction system.The object-oriented method was used to analyze and design the system,and the main functions of the system were realized.Merchants can use this system to personalize the delivery of coupons,which can reduce marketing costs for merchants and improve operating efficiency.Compared with other models,O2O coupons usage prediction models based on LightGBM has higher accuracy and faster training speed.The O2 O coupon usage prediction system mainly includes a data calculation module,a model prediction module and a Web server module.The data calculation module is implemented by Spark.Its main responsibility is to calculate related features,including user-related features,merchant-related features,coupon-related features,and so on.The model prediction module is implemented in Python,and its main responsibility is to call the trained machine learning model according to the characteristics of Spark calculation to get the prediction result.The Web server module is implemented based on the SSM framework.Its main responsibility is to provide external services and interfaces for web page display,including interfaces for viewing forecast results,viewing consumption records,and using coupons..Through this system,it is possible to predict the consumption probability of consumers using coupons,and then to personalize the delivery of coupons,giving merchants strongermarketing capabilities.