In this paper, we first preprocessed the user’s shopping behavior data, set the prediction goal, constructed the features of the user’s online purchasing behavior prediction model, and classified and selected the constructed features based on the SVM-RFE algorithm. Then, on the basis of the fuzzy neural network algorithm of fuzzy theory, the network purchasing behavior prediction model was constructed by combining the assessment indexes of the prediction model results as well as the 5-fold cross-validation method. Finally, the evaluation results of the prediction model are examined and compared with common prediction algorithms to confirm the performance of the algorithm in this paper. The results show that the average relative error of model training can reach 0.013, and the absolute error with the actual value ranges between [0.01, 0.06]. On the same test set, the F1 value of the prediction model in this paper is between [0.88, 0.91], and the F1 value of the algorithm on each test set has a small difference of only 0.03, and the F1 value of the other prediction models has a maximum difference of 0.09. The prediction model constructed in this paper has a good prediction effect and robustness.