| With the progress and development of modern science and technology and society,the information management system has replaced the previous way of text record management,especially the management upgrade in the field of education.The behavior data of teachers and students in campus has been recorded more and more by the Internet of things technology,laying a data foundation for the realization of digital management of campus.Aims to use data mining analysis of mathematical statistics,machine learning and computer technology,in-depth analysis of the hidden education data after the valuable information,thus realize teacher to student’s individualized education,promote students’ learning results,help students comprehensive development,optimize the educational teaching design,improve school management level.To realize the improvement of these aspects is inseparable from the analysis of students’ learning situation,so it is particularly important to build a set of more general learning situation analysis and prediction system to improve the overall information level in the field of education.Learning in the analysis precision of student achievement is also very important to control and predict,accurate performance prediction helps teacher to student learning situation to make correction in time,help the school make targeted optimization of teaching design in time,help parents understand students’ learning trend in time,at the same time help students for individual development plan in time.In student performance prediction,a random search algorithm is proposed for hyperparameter optimization of Stacking models.Compared with grid search algorithm,random search algorithm can select suitable hyperparameters for the model more efficiently.Compared with the ridge regression,SVR,random forest,and Adaboost data mining algorithms,the optimized RG-UNNS and RG-CNNS fusion algorithms significantly improved the accuracy of student performance prediction,and reduced the MSE by 18.2% and 10.57%,respectively,compared with the linear Stacking fusion algorithm.Rg-unns has the best prediction effect when the time requirement is not high.In the aspect of system construction,the study situation analysis and prediction system for senior high school is constructed by using B/S three-layer architecture mode.Combining students information,subject information,multi-purpose card consumption data,attendance information,performance information,such as learning data,for an overview of the different scene needs to build a school,students basic information,academic performance,personal attendance and IC card consumption,meet a portrait,class,group analysis module of the analysis of the learning data statistics,Through a variety of easy-to-observe charts for visual display,to provide users with simple and convenient learning situation analysis services. |
