Research And Implementation Of College Student Academic Warning System Based On GA＿XGBoost

With the rapid development of internet technology and artificial intelligence,the management of education informatization has made a leap forward.Nowadays,major universities have begun to adapt to the needs of the information age,actively carrying out corresponding informatization construction,and establishing their own information centers.Unfortunately,the massive student data stored in these information center databases has not received enough attention and utilization,and the value they contain is often overlooked.However,in the context of the rapid development of the Internet,these data have gradually attracted the attention of researchers.How to identify students who may face academic failure risks by using these data,in order to reduce academic failure in higher education and improve the academic warning and assistance mechanism,has become a highly concerned topic.In the current field of higher education,warning and identifying students’ academic risks are of great significance.Academic warning is a key technology in student academic warning systems.In this thesis,we constructed an efficient and accurate student academic warning model based on the GA＿XGBoost algorithm,which can quickly and accurately evaluate the academic status of students.On this basis,we designed and implemented a student academic warning system,which can help schools to timely identify students’ academic problems and provide effective intervention measures to improve students’ academic performance and help them successfully complete their studies.The specific research work is as follows:1.Handling of Student Data: Unlike the typical approach of limiting student academic warning results to just success or failure,this thesis introduces a third intermediate category and divides student academic warning results into three categories(unfinished studies,extended completion of studies,and smooth completion of studies).At the same time,student data is preprocessed,which mainly includes data cleaning,feature selection,data transformation,and imbalanced data processing.2.Constructing a Student Academic Warning Model: Based on the XGBoost algorithm model training,this thesis introduces the genetic algorithm(GA)and constructs a student academic warning model based on GA＿XGBoost.At the same time,using accuracy(Accuracy)and F1 score as evaluation indicators,GA＿XGBoost student academic warning model is compared with XGBoost,SVM,RF and other algorithm models,demonstrating the outstanding performance of the warning model constructed in this thesis in corresponding model evaluation indicators.Combined with the SHAP framework,the interpretability of the GA＿XGBoost student academic warning model is analyzed.Firstly,student feature importance is sorted using SHAP,and the dependency relationship between student features and the impact of this dependency relationship on academic warning results are analyzed.Secondly,combined with the individual student,their academic warning results are analyzed specifically to provide assistance for their academic performance.3.Building a student academic warning system: Firstly,the functional and nonfunctional requirements of the student academic warning system were analyzed.Then,the system design was described in detail using diagrams.Using the Express and React frameworks,the implementation of the GA＿XGBoost student academic warning system was completed.The user roles of the system are administrators,education managers,and students.The main functions of education managers include viewing student academic warning information and proposing targeted strategies.The main functions of students include managing personal information,viewing academic warning results,and viewing evaluations from education managers.After students complete their information input,education managers use the academic warning function to issue academic warnings to students.Finally,the system was tested,and the results showed that the system has a good warning effect on student academic warning and can provide reference value for subsequent related research.