Light GBM and GA Based Algorithm for Predicting and Improving Students' Performance in Higher Education in the Context of Big Data

The emergence of artificial intelligence has improved human life, and it has been extensively applied in healthcare, finance, education, transportation, etc. To predict and enhance the academic performance of college students and promote the improvement of college education methods, the study constructed an academic performance prediction model on the ground of lightweight gradient boosting machine, designed a model of academic performance improvement strategy on the basis of genetic algorithm, and collected the academic performance data of college students. In addition, the study also designed the fitness function of the genetic algorithm. The results show that the accuracy and the maximum root mean square error are 98.71% and 8.71, respectively. The performance enhancement strategy model tends to level off after nearly six iterations, and the minimum value of the loss value is 0.13, and the maximum value of the mean absolute error is 6.21.In the practical application, the predicted total grades of nearly 315 students out of 320 students and their real total grades were consistent, and the students' online learning grades, midterm exam grades, regular grades, and overall grade point average were upscaled by 7, 10, 5, and 9 points, respectively. The research design of the performance prediction model and performance improvement strategy model has good performance, providing support for performance prediction and improvement to enhance the quality of higher education.