Research On Early Warning Methods Of College Students' Achievement Based On Integrated Learning

Academic early warning is an important means to strengthen school management,improve students' self-management and self-discipline,as well as form a common management model for schools and parents.The practice of implementing a number of colleges and universities early warning system proves that through the development of academic early warning work,schools,parents and students can work together to promote students' correct attitude towards of learning,study hard,and successfully complete their studies.To this end,how to use the various information implicit in the learning,living data and academic performance of college students,to establish a more scientific and accurate academic early warning model,and to help students avoid problems in the learning process,not only caused the education sector It has been widely concerned and has become a hot and difficult issue in data mining and artificial intelligence research in the field of education.Based on the above background,in recent years,scholars at home and abroad have carried out a large number of researches from various aspects such as student behavior patterns,comprehensive evaluations,and early warnings has achieved certain research results.However,in general,due to the complexity of different majors and academic influencing factors,there is still has much space for development in the study of college students' early warning models.In view of this,based on the study of existing research results,this paper has carried out related research on the issue of academic early warning for college students,from the curriculum relationship network and student behavior model.The main research contents are as follows:(1)By collecting various behavior data of college students in campus activities,including consumption data,access control data,Internet data,etc.,a large data set of student behavior was constructed.On this basis,the analysis determines the various behavior patterns behind the student behavior data and the main influencing factors related to the student's academic performance.The feature selection and feature extraction of students'behavior data are carried out from multiple perspectives.(2)A weighted Apriori algorithm based on vector matrix optimization self-connect is proposed.The algorithm is based on the traditional Apriori algorithm repeatedly scanning the database,the self-connecting efficiency is low,etc.The vector matrix is used to optimize the elf-connecting calculation method,and the effective weighted association rules are more efficiently mined,which avoids the invalid and uninteresting association results.And apply it to the construction of student relationship network and the mining of management rules.The experimental verification proves the effectiveness of the proposed method for constructing the curriculum relationship network and mining the association rules of different academic achievements.(3)Combining the extracted student behavior characteristics and the constructed course relationship network,using the decision tree,NB,SVM and other algorithms to establish a variety of early warning models of course performance and comparative analysis.On this basis,a method of constructing college students' early warning model based on Stacking integrated learning is proposed.Using this method,the student performance warning model is obtained by selecting all the course performance data,life data and online record data of a university with a total of 275 students in a certain period from 2009 to 2012 as the experimental data,and the empirical prediction is 2013.Among the academic achievements,people have achieved good prediction results,which proves that the proposed method can effectively improve the accuracy of college students' early warning.