Research And Application Of Process-oriented Student Comprehensive Ability Assessment

The flourishing development of artificial intelligence has promoted the modernization process of educational assessment reform,and algorithms such as machine learning are applied to the field of educational assessment to solve potential problems in the current stage of educational assessment,such as teachers’ inability to accurately perceive students’ understanding of the curriculum and students’ inability to self-assess their mastery of knowledge.As the most basic kind of educational evaluation,several studies in recent years have been devoted to improving the accuracy of predicting learning performance for better assessment,but the interpretability of assessment models is not strong.Comprehensive ability assessment of students is a key aspect in educational evaluation.Most studies assess students’ comprehensive ability by collecting learning-related data sets,ignoring information on dimensions such as ideology and morality,physical and mental health,and practical ability,etc.Moreover,the assessment results of most schools for students are mainly based on grade ranking display,ignoring students’ process performance in daily life,and the final obtained evaluation results are not applicable with the current The final evaluation results are not applicable to the current comprehensive talent cultivation system.In order to evaluate students’ comprehensive quality more accurately and cultivate all-round students who can better meet the needs of society,this paper conducts research in the following three aspects to address the problems of the above-mentioned studies.(1)To address the problem of poor interpretability of assessment models,a learning achievement assessment method integrating Light GBM and SHAP is proposed.By using the Light GBM algorithm to construct a student achievement level assessment model,the accuracy of the model is improved by 2.2% compared with existing assessment models,and SHAP to interpret and analyze the constructed student achievement level assessment model,the interpretability of the model is enhanced while the assessment accuracy is improved.(2)To address the problems of single source of data and inaccurate assessment of existing students’ comprehensive ability level assessment,a comprehensive ability assessment model for college students based on IAHP-BP neural network is proposed.The IAHP algorithm is used to determine the assessment indexes of multiple dimensions of students’ comprehensive ability,and combined with BP neural network to assess comprehensive ability,combining qualitative and quantitative analysis,and limiting the relative error to less than 2%,which makes the assessment results more reliable.(3)A process-oriented assessment system of students’ comprehensive ability was developed.Through the statistical analysis of students’ process data,students’ recent academic life performance can be viewed dynamically.The system visualizes students’ assessment results in a multidimensional display,and both students and teachers can understand the current learning situation and learning quality from the visualized results in a timely manner,so as to explore the preferential strategies for students’ learning and working paths.