摘要
达成度评价是否科学是研究人员关注的重要问题,把机器学习技术用于优化和完善目标达成计算有助于更加科学地评价目标达成情况。针对当前目标达成评价方法和对特异数据处理的不足,以教学目标达成度为研究对象,研究了目标达成度的模糊性、指标量纲差异性、偶然性等主要特征,提出贝叶斯学习和马氏距离相融合的目标达成计算模型及算法。通过计算样本的均值、方差、协方差矩阵、协方差矩阵逆矩阵等过程,计算出样本与目标之间的马氏距离,用马氏距离的平均值作为阈值,得到样本目标正反例;用贝叶斯概念学习算法,计算样本为正反例的贝叶斯概率,根据概率平均值阈值,综合马氏距离和贝叶斯概率,得到目标达成的分类结果。用教学中的数据进行测试,结果表明,贝叶斯概念学习结果与马氏距离相融合,能发现和处理特异数据,有助于使目标达成结果更加科学。
Objective achievement evaluation based AI is widely used in areas of engineering management,education and teaching management,risk prediction and evaluation.Teaching evaluation is a main problem for many educators.Because of the limitation of relying on a single weighted average in the current process of achieving educational evaluation goals,we studied on the achievement of course goals,focused on the main characteristics of achievement,and combined them with the characteristics,Proposed a target achievement calculation method based on Mahalanobis distance similarity measured.By calculating the mean,variance,covariance matrix,and inverse covariance matrix of the samples,the Mahalanobis distance between the samples and the target is calculated,and the target achievement status is measured based on this distance.This method considers the factors such as the ambiguity of achievement degree,the objectivity of achievement target indicators,and the differences in sample indicator characteristic factors,which can better handle multidimensional data,and also,we discussed the calculation process of course goal achievement degree.
作者
王刚
WANG Gang(School of Education,Ankang University,Ankang 725000,Shaanxi,China)
出处
《安康学院学报》
2025年第6期79-83,95,共6页
Journal of Ankang University
基金
2023年陕西省高等教育教学改革项目“专业认证背景下小学教育专业特色打造策略研究”(23BY155)。
关键词
机器学习
马氏距离
贝叶斯概率
达成度
教学管理
machine learning
Mahalanobis distance
Bayesian probability
goal achievement degree
teaching management
