In this paper, we use the real data provided by Tianchi Big Data Research Platform to predict which commodities will be purchased by this user in the short term among the commodities that the user has interacted with. Firstly, the collected historical interaction data of commodities are normalized by Z-score. Then the features are extracted for coding, and the number of features is reduced based on the chi-square test method to improve the modeling efficiency and accuracy. Finally, the processed user purchase behavior data is input into the logistic regression model for e-commerce user purchase behavior prediction. The AUC value of the logistic regression model is greater than 0.5, the percentage of the number of purchasers increases with the increase of the predicted probability value, and the number of non-purchasers decreases with the increase of the probability value of the score segment. The model prediction results are consistent with the actual purchase, and the model is valid.
