E-Commerce User Purchase Prediction Based on Improved Machine Learning Algorithms

The role of consumer shopping demand in the operation forecast of e-commerce platforms is very important, but there is the problem of incomplete data estimation. The k-NN nearest neighbor algorithm cannot accurately predict the problem of user purchasing power in e-commerce platform data, and the sample is unbalanced, and the problem is handled in a single way. Therefore, this paper proposes a machine learning algorithm to analyze the shopping needs of e-commerce enterprises. Firstly, the Markov chain theory is used to evaluate the purchase behavior of users, and indicators are carried out according to the requirements of purchase historical data Divide and reduce distractions in purchasing behavior. Then, Markov chain theory evaluates the purchasing power of consumers, forms a consumer shopping demand program, and has a consumer shopping demand The results were comprehensively analyzed. MATLAB simulation shows that under certain evaluation criteria, the machine learning algorithm is accurate to the purchasing power of e-commerce platform users The range of consumer demand is better than the k-NN nearest neighbor algorithm.