A Study Of Academic Early Warning In Undergraduate Big Data Professional Training

People-oriented and student-oriented is the core educational philosophy of modern university teaching.In the era of rapid modernization,using advanced science and technology to solve students' academic problems is an important way to implement this philosophy.Based on multiple perspectives,making full use of the available data to warn students of abnormal situations in advance and using the obtained information for teaching management is now a hot topic in universities.In this paper,we first analyze quantitatively the course performance data in teaching practice with the help of correlation data mining,and get the significantly correlated courses through simple correlation analysis,and further analyze and explain the significantly correlated courses,followed by using Apriori algorithm to find the connection between prerequisite courses and post-requisite courses starting from mining the correlation between course knowledge.Based on this finding,students with unsatisfactory prerequisite courses were alerted.In order to predict students' grades for the next semester,a common classification prediction was first used,and the collected data was combined with student questionnaires to perform a more comprehensive pre-processing work on the data through steps such as data cleaning,selection,and so on.With the help of classification algorithms of logistic regression,neural network,and support vector machine,which are commonly used in machine learning,a three-classification prediction model for grades was constructed.The prediction performance of the model is demonstrated with the help of images,and the evaluation of the experimental results is given,showing that the support vector machine has high accuracy in prediction.In order to further improve the experiment,the data was firstly enhanced for the small amount of data,and the predicted values of the next semester's grades were output by applying CNN+LSTM to predict the total grades of the next semester considering that the students' grades are a continuous dynamic development process.In order to achieve the effect of early warning for the class collective.Finally,in order to implement academic early warning specifically to individuals,an outlier detection model is proposed to identify some special learners starting from outlier detection,and special outlier students are found and their academics and behaviors are analyzed for the purpose of targeted academic guidance as well as early warning.The paper uses actual data of students in school to build a prediction model of student performance from different aspects,which enables teaching and learning to study students more specifically and comprehensively based on the results,thus achieving early warning.