Application Research Of Deep Forest Algorithms Oriented To Commodity Recommendation

In recent years,online transactions have sprung up,and online trading platforms have permeated all aspects of people’s lives.While e-commerce brings convenience to people,because it covers a wide range of commodities and diversified marketing methods,these complex information brings many inconveniences to users when selecting commodities.It excavates potential and valuable customer behavior data from massive web data,provides reliable technical support for users’choice and business operation,and improves user experience.It is of great significance to realize mutual benefit and win-win between businesses and users.The main research work of this paper is as follows:(1)At present,the research on user’s operation behavior prediction methods usually only focuses on one kind of user’s operation behavior,which can not fully reflect the overall characteristics of user’s behavior.Therefore,through the technical means of in-depth mining and analysis of user behavior,through the construction of user behavior characteristics engineering,establish the overall user behavior characteristics model.By learning the theory of decision tree and ensemble learning,a prediction model of user purchase behavior based on deep forest is proposed,and the structure chart of the model prediction is designed in the context of e-commerce.(2)In the context of electric shopping malls,it is necessary to predict whether users have purchasing behavior or not.However,the number of purchasing behavior is relatively small compared with other operations,which will lead to category imbalance.Therefore,different methods of over-sampling or under-sampling are used to process the data,so that the samples of category balance can be obtained before the input model.The deep neural network algorithm is used in the data set to compare its performance with the traditional machine learning method and the model proposed in this paper.(3)On the basis of the research work in this paper,a prediction model of user purchasing behavior based on deep forest is constructed from data processing,feature construction,model selection,model optimization and result prediction.The detailed design and implementation process of the model are given,and the F1 value is used as a measurement index.The experimental results show that the model reduces the time cost,but also reduces the time cost.The prediction accuracy is improved.