The evaluation of teacher’s teaching ability and student online learning prediction based on optimized cluster analysis and machine learning

In the past, the evaluation of teacher’s teaching ability mainly relied on qualitative evaluation methods, often influenced by subjective factors. Nowadays, student evaluation of teaching is an essential means of monitoring teaching quality, as well as a right and obligation of students. To accurately obtain teaching evaluation results, evaluate the teaching ability, and improve teaching quality, clustering analysis of teaching evaluation data is studied. It adopts an improved K-modes clustering and a neural network prediction model based on machine learning to predict the effectiveness of online learning for students. It conducts experiments to verify it. Experimental studies have shown that the absolute error of training results is 54% in the range of 0–1, and 7% in the range of greater than 3. The similarity between the initial cluster centers selected by the random method and the cluster centers of the sample data is unstable, while the cluster centers from minimizing the sum of squared errors are closer to the sample data. The clustering effect of removing abnormal data is more concentrated. The accuracy of the clustering center distance calculation method studied is 0.9816, while other methods are 0.9321 and 0.9287. The calculated cluster center distances are stable, including values of 0.8307, 0.8547, and 0.8325. It showcases that research methods and predictive models can optimize the evaluation of higher education. Through student evaluation of teaching, it is possible to promote teaching and learning through evaluation, strengthen university teaching, and serve the national economy and society.