The Behavior Analysis and Achievement Prediction Research of College Students Based on XGBoost Gradient Lifting Decision Tree Algorithm

Abstract

Student behavior data is a reflection of students' learning styles and habits in campus. Quantitative evaluation of various behavior indicators of students' college learning life can reflect students' learning status and laws, and the academic performance is the most important and basic index to evaluate a student's learning situation. Based on the data of students' behavior under the Four PIN education system of Beihang Shoue College, this paper adopts XGBoost gradient upgrade decision tree algorithm to fully mine and analyze the situation of college students' study life and participation in social work, and to study the potential behavior patterns with strong correlation between students' behavior data and students' performance. The performance prediction model is established through data mining technology and the prediction accuracy is 73%. The study finds that Pinjian program records the growth of students. Students' overall evaluation and student performance shows a positive correlation. Pinxue program promotes academic development. Academic guidance in college plays an important role in improving students' academic performance. Pinzhi program enhances comprehensive literacy. Cultural and artistic activities and serving as student cadre show a positive correlation with academic performance.