Modelling And Application Of Student Learning Behavioural Characteristics In School

The core priorities of national education development are the continuous improvement of teaching,learning,and quality talent training.With the development of information technology,the amount of data in various fields has exploded,including education.A significant portion of educational data is comprised of student examination results,book lending information,consumption records,and attendance records.How to use the growing amount of educational data to help teachers improve their management skills and efficiency,to help students self-examine,and improve their self-learning skills are the main problems that need to be solved in smart campuses now.In the face of huge education data,relying only on traditional processing methods will inevitably cause problems,such as teachers’ increased management pressure and some student’s inability to obtain timely guidance.Therefore,this paper constructs a student learning behavior analysis model,aiming to use modern information technology to promote the development of education and teaching,and improve learning efficiency and learning effectiveness,with the following specific work:(1)A hierarchical multi-label classification algorithm is proposed.In response to the current problems of computationally intensive multi-label classification problems,easily broken inter-label relationships,and weak interpretability,this paper proposes a hierarchical multi-label classification method based on cluster-like relationships.The algorithm first uses improved K-means to cluster the labels in the sample set multiple times to form clusters with hierarchical relationships;then selects base classifiers to train the data according to the labels in the clusters,which can not only reduce the training cost but also retain the relevance of the labels;on this basis,the base classifiers are combined into multi-label cluster clustering trees,and then uses the integration idea to construct the cluster clustering trees into random The algorithm’s performance and generalization ability are improved by combining the base classifiers into multi-label cluster trees and then constructing the cluster trees into random forest classification models using the integration idea.The experimental results show that the algorithm’s overall performance outperforms other algorithms,especially in larger datasets,where all the metrics rank first.(2)An extreme random forest classification method is proposed.The extreme random tree aims to improve the generalization ability of the algorithm by increasing the randomness,while the randomness of the base classifier(decision tree)is increased,which must lead to a decrease in the classification effect.This paper proposes an adaptive scale extreme random forest classification method based on the above characteristics.The algorithm first selects a voting mechanism associated with accuracy to enhance the fitting ability;then proposes an adaptive parameter selection method to explore the optimal combination of parameters starting with default parameters to balance the relationship between algorithm bias(Bias)and variance(Variance).The experimental results illustrate that the algorithm has better performance in both classification effect and generalization ability.(3)Construction of a learning behavior analysis system.According to the characteristics of student learning behavior data and the research needs,a model of student learning behavior analysis was formed by combining the two algorithms proposed in the paper,and a learning behavior analysis system was constructed based on this model.In this system,through the analysis of database data,students’ personality labeling is completed,students’ learning status and learning ability are displayed,and factors with greater influence are explored to form a result report,which provides reliable data support for teachers’ management and students’ learning.