User Identification Algorithm in E-Commerce Based on Support Vector Machines And Clustering Detection

Existing e-commerce systems have the problems of large data volume and many users, which cannot effectively identify high potential users. To address this challenge, this paper proposes an e-commerce high potential user identification algorithm based on support vector machine with clustering detection. The algorithm aims to address the problem of classifying e-commerce customer information. The algorithm modifies the support vector machine by proposing the adopted kernel function, which undergoes initial training with the standard support vector machine and then retraining with the new vector machine. The algorithm is trained by introducing the error function F to obtain more reliable information. At the same time, a stabilisation of the data set is achieved by combining clustering detection, which in turn makes the algorithm more adaptive. Experimental results show that the algorithm significantly improves the processing speed and accuracy of user classification, helping companies to develop more accurate marketing strategies. On specific data, our algorithm improves recognition accuracy by $4 \%$ compared to traditional methods. This algorithm provides new ideas and methods for identifying high-potential users in e-commerce.