Research On Consumers’ Willingness To Buy Online Based On Machine Learning

With the advent of the era of big data,more and more customers tend to online shopping network stores by way of the consumption quantity increased,the electricity industry increasingly fierce competition in order to better help the electricity industry,enterprises should conform to the development of The Times,combining the development trend of current and future industry another road,find the current electricity industry business model electric business platform for consumers when shopping to browse,leaves a lot of interactive data,for the enterprise marketing strategy formulation in the future,provides a solid foundation.Consumers’ online shopping intention in research,based on machine learning as the background,through the way of XGBoost + logistic regression models,to discuss consumer’s online shopping intention in this paper,on the data obtained from the first of a series of data preprocessing,and then the processed data,carried on the characteristics of engineering data is established and divided into training set,the original data set is constructed according to 7 kinds of time granularity and testing set again XGBoost model was determined by the parameter tuning of the optimal parameters,and the parameter tuning of the AUC inspection result,ultimate AUC value reached 97.78%.Furthermore,a visual demonstration of the XGBoost+ logistic regression model was carried out.Finally,the values of accuracy and recall rate were determined by the evaluation function.Finally,the accuracy of prediction of consumers’ online shopping intention was 80.7% through the calculation of F1 value.To the importance of the characteristics in the process of building the model output,the calculated results are comparative analysis,the following conclusions: When purchasing goods,consumers are mainly affected by negative rating rate,brand,browsing behavior,clicking behavior,being added to the shopping cart,deleting behavior and so on.Therefore,as an e-commerce platform,it should actively and rationally use the interaction records left by users in the e-commerce platform,and better find potential users according to the behavior records of users,so as not to lose users.