Research On Student Early Warning Model Based On Machine Learning

Education is one of the important tasks of national development,and it is the necessary way for the country and society to cultivate high-quality talents.The academic level of students is an important indicator of the quality of college education and teaching,the effectiveness of teachers' teaching and the learning ability of students.To improve students' academic level,not only the enhancement of the teaching level but also advanced technology and methods according to specific demand are needed.We need to accelerate the pace of improving students' learning quality and academic level as well as school teaching quality from multiple directions.In order to improve students' academic level from the aspects of technology and methods,an academic warning model based on student mark is put forward in this paper.This academic early warning model analyzes and calculates the statistics of a university student's achievement to realize the prediction of students' courses and graduation,so as to achieve abnormal warning of student mark.In the research of student course prediction,the core of the model is to train the course warning model using the Bagging-C4.5 decision tree algorithm and optimize the model using genetic algorithms.In the study of student graduation prediction,the basic idea is to use the weighted Naive Bayes and C4.5 decision tree algorithms to build the student graduation prediction model based on the collected student data.When conducting prediction research on students' courses,the model first use association rules to mine courses that have a greater impact on the predicted courses,and use them with other features to construct experimental data sets.Then the Bagging-C4.5 is used to train samples to obtain the classification model.Since the Bagging-C4.5 algorithm trains the model to produce the same or similar C4.5 base classifier,genetic algorithm is used to search for the best base classifier for integration.Then the classification effect and running speed of the model are improved.When conducting prediction research on student graduation,the model first use Naive Bayes algorithm and C4.5 decision tree algorithm to train the classification model.When using Naive Bayes algorithm to train the classification model,information gain is used to weight the attributes to improve its classification.Then the two models are combined according to a certain strategy to obtain a model for predicting student graduation.Compared with the existed algorithm,the method adopted in this academic early warning model in terms of course prediction and graduation prediction has higher accuracy and more accurate prediction results.The academic early warning model has higher performance.