College English speaking, as an important part of English teaching, is getting more and more attention from many scholars. Data mining technology has brought new opportunities for teaching spoken English in universities. The article builds a portrait and analyzes students through the data generated in the process of teaching spoken English at university in order to understand their relevant attributes and learning habits. Then, the decision tree model is used to predict several scenarios of students’ oral English performance. Finally, 2 classes of a university majoring in English were selected as the empirical research subjects to be tested. In the analysis of the changes in the oral English scores, the mean score of the pre-test of the control class was 21.43. The mean score of the pre-test of the experimental class was 21.29, which led to a small difference in the mean scores of the pre-tests of the two classes. In contrast, after the experiment, the mean score of the post-test of the control class was 27.51, and the mean score of the post-test of the experimental class was 30.85. The experimental class’s mean score increased by 3.34 points, and the overall English-speaking performance saw a more significant improvement.