Research On Database Marketing Of Lister Tmall Flagship Store Based On Machine Learning Algorithm

The human technological revolution has subverted the perception of small merchants and hawkers who only knew about stall trading more than a decade ago.From simple buying and selling relationships to complicated online transactions at this stage,it not only highlights the advancement of technology and the times,but also indirectly illustrates the diversified evolution of marketing methods and means.In the domestic and foreign marketing markets,database marketing has been favored by more marketers due to its low-cost and efficient marketing methods,and has become a "rookie" in the marketing market in recent years.However,as a prophet and ferry in the unpredictable market,database marketing tends to be more applied in simple practice after simple data analysis.In contrast,database marketing combined with machine learning algorithms has always been a depression in the marketing field.Based on this,this article takes the Lister Tmall flagship store as the research object,mainly from the four stages of theoretical preparation-status analysis-model construction-results application based on machine learning algorithms for database marketing research.First,it discusses the connotation,advantages and specific operating proceduresof database marketing.At the same time,four excellent machine learning algorithms with logistic regression,random forest,support vector machine,and GBDT are selected to elaborate from two perspectives,namely the basic principles and algorithm introduction,to lay the theoretical foundation for the model training chapter.Secondly,analyze the current status of Lister Tmall flagship store in terms of distribution of marketing objects,proportion of marketing channels,composition of marketing methods,etc.,and find new marketing ideas for database marketing using machine learning algorithms based on the main problems at this stage.Thirdly,under the premise of the marketing ideas,data acquisition-data processing-positive and negative sample settings.At the same time,based on four machine learning algorithms,features are combined from three perspectives: consumer,store,and consumer-store relationship.Due to the different parameter settings of each model,a total of 38 models and 4 machine learning algorithms are used to build 12 models for training,and the evaluation indicators are used to compare the output effects of each model.The best-effect GBDT3 is selected as the prediction model.Finally,by substituting the prediction samples into the GBDT3 model for testing,a group with a prediction score in the range of80-99 was selected as the model prediction group to put on the market,and the Lister Tmall flagship store should be based on the model prediction database marketing method.Because on the one hand,usingmachine learning algorithms can solve the problem of uneven distribution of marketing objects,on the other hand,it can effectively prevent the loss of potential consumers.In addition,it also proposes application strategies that optimize other database marketing methods and assist model prediction to improve marketing effectiveness.