Research on the Influencing Factors of E-commerce Conversion Rate Based on Regression Analysis and CART Decision Tree Algorithm in the Context of Big Data

Abstract

The issue of enhancing the platform conversion rate has emerged as a primary challenge hindering the advancement of e-commerce platforms. This study undertakes a comprehensive analysis of the interplay between the platform conversion rate and its influencing factors. To this end, a two-pronged approach is adopted, employing both a polynomial fitting model and a decision tree model to elucidate the underlying mechanisms. Initially, a thorough data cleansing and descriptive analysis is conducted on the voluminous data set provided by Alibaba Tianchi Lab. Subsequently, a polynomial fitting model is formulated. The random forest feature importance analysis revealed that the total number of user visits is the primary factor influencing the platform conversion rate. This variable was then employed to construct a polynomial fitting model based on the least squares method, which was found to have a strong positive correlation with the platform conversion rate. Finally, the decision tree model was established. The final decision tree model is obtained by using CCP pruning after establishing the decision tree model with depth restriction. The study shows that the conversion rate of the e-commerce platform is mainly related to user browsing, user age, and user attributes. The results of this research can provide a scientific basis for e-commerce enterprises to accurately formulate marketing strategies and optimize the operation process, which can strongly promote the efficient development of the e-commerce industry.