随着大数据技术的兴起，利用校园卡数据、学生管理系统数据的教育统计研究较为常见。然而，有关学业成绩的实证分析、得出定量结论的研究并不多见。为此，本文采用固定效应模型、多元线性回归模型进行研究，基于图书馆借阅行为数据、校园网使用数据、体育锻炼数据及校园卡消费数据等日常行为数据对大学生学业成绩进行预测研究，通过探究日常行为对学业成绩的影响因素，建立有效的成绩预测模型。本文通过描述性统计进行初步观测，构建固定效应模型、多元线性回归模型，模型均能够达到比较准确预测学生学业成绩的目的，并进行模型比较，分析不同模型的异同，并提出建议，以促进大学生有效提高学习效果，并为学生提供学习习惯培养的建议。

With the development of big data technology, researches utilizing campus card data and student management system data for educational statistics have become common. However, empirical analysis and quantitative conclusions regarding academic performance are not frequently seen. Therefore, this study employs fixed effects models and multiple linear regression models to investigate and predict college students' academic performance based on daily behavioral data such as library borrowing behavior data, campus network usage data, physical exercise data, and campus card consumption data. By exploring the influencing factors of daily behaviors on academic performance, effective performance prediction models are established. This study conducts preliminary observations through descriptive statistics, constructs fixed effects models and multiple linear regression models, both of which can accurately predict students' academic performance, and compares different models to analyze similarities and differences. Suggestions are provided to promote effective improvement of learning outcomes for college students and to offer advice on cultivating study habits.