This study proposes a strategy for predicting English listening learning performance and English listening achievement and constructs a corresponding model, which provides a basis for proposing a personalized cultivation pathway for English listening by implementing an English listening prediction model. The learners’ memory forgetting and learning behaviors are integrated into the Bayesian knowledge tracking model to obtain the BF-BKT model, and its corresponding CPT and Bayesian formula is updated to realize the prediction of English listening learning performance. With the help of the K-means algorithm to mine the relationship between learning behavior and learning performance, the gradient progressive regression tree in deep learning is used to analyze the connection between key influencing factors and performance, and the English listening learning performance is predicted. In this paper, the English listening learning performance prediction model ROC area AUC=0.826, and the relative error of the performance prediction model is always lower than that of the traditional model, with better prediction performance. Applying the personalized English listening cultivation pathway based on English listening prediction for English listening teaching, the total listening score of the experimental class was 36.88, which was higher than that of the control class by 5.62, showing a difference (P<0.01). The mean value of each specific strategy in English listening learning strategies was higher than that of the control class, and the overall listening teaching satisfaction reached 3.87, which was greater than that of the control class by 0.66.