Student Learning Status Prediction Based on RFECV-PSO-LightGBM in Colleges

Due to the implementation of information management in colleges, a large number of data related to student behavior are generated. How to combine the data of students' daily life to predict and classify the students as is a challenge faced by administrators of colleges at present. The paper constructs the characteristics matrix of students' learning situation, through statistical analysis of various data generated by students' daily behavior. RFECV was used to extract important features and the best number of features, and LightGBM was used to predict students' total scores. PSO algorithm was used to optimize the prediction model parameters, and the PSO-LIGHTGBM model had the best fitting effect after comparing the experimental data. In the paper, the regression model based on RFECV-PSO-LightGBM algorithm is proposed to solve the problems of difficult and complicated in student learning prediction. Improve the quality of education and teaching.