<abstract xmlns="http://www.w3.org/1999/xhtml">

<p>This paper collates and fuses data from various aspects of university students and designs student evaluation indexes by combining the data. Based on the student behavior segmentation model of cluster analysis, the traditional k-means clustering algorithm is optimized, focusing on the optimization of content selection of initial cluster centers and distance optimization calculation to make a better clustering effect and classify students accurately. Then the K-nearest neighbor algorithm is proposed to solve the problem of timeliness and posteriority of student warning, by which the behavior of students can be predicted, and teachers can manage and help students in time. Finally, a student behavior characteristic category model is established to strengthen the prediction model of the K-nearest neighbor algorithm and improve the accuracy of predicting student behavior. The student data of a university were analyzed, in which the correlation pearson coefficient between the number of library admissions and grades ranged from 0.25-0.31, and the p-values were all much less than 1. It showed that the number of library admissions of students obviously affected students’ grades. In the analysis of the correlation between the early rising index and academic achievement, where the Pearson coefficient is 0.291 with a p-value of 5.61E-7. This indicator, which is not much associated with academic behavior, indicates that students’ early rising and timely breakfast can affect achievement to some extent. Therefore, the method of this paper can be refined using student data.</p>
</abstract>

This paper collates and fuses data from various aspects of university students and designs student evaluation indexes by combining the data. Based on the student behavior segmentation model of cluster analysis, the traditional k-means clustering algorithm is optimized, focusing on the optimization of content selection of initial cluster centers and distance optimization calculation to make a better clustering effect and classify students accurately. Then the K-nearest neighbor algorithm is proposed to solve the problem of timeliness and posteriority of student warning, by which the behavior of students can be predicted, and teachers can manage and help students in time. Finally, a student behavior characteristic category model is established to strengthen the prediction model of the K-nearest neighbor algorithm and improve the accuracy of predicting student behavior. The student data of a university were analyzed, in which the correlation pearson coefficient between the number of library admissions and grades ranged from 0.25-0.31, and the p-values were all much less than 1. It showed that the number of library admissions of students obviously affected students’ grades. In the analysis of the correlation between the early rising index and academic achievement, where the Pearson coefficient is 0.291 with a p-value of 5.61E-7. This indicator, which is not much associated with academic behavior, indicates that students’ early rising and timely breakfast can affect achievement to some extent. Therefore, the method of this paper can be refined using student data.

An analysis of the innovation strategy of student management in universities based on big data

Applied Technology College of Soochow University

Applied Mathematics and Nonlinear Sciences

This work is licensed under the Creative Commons Attribution 4.0 International License.

This paper collates and fuses data from various aspects of university students and designs student evaluation indexes by combining the data. Based on the...

An analysis of the innovation strategy of student management in universities based on big data

Publié en ligne: 02 oct. 2023

Pages: -

Reçu: 18 oct. 2022

Accepté: 12 avr. 2023

DOI: https://doi.org/10.2478/amns.2023.2.00443

Mots clés
Student management, Behavior prediction, K-means clustering algorithm, K-nearest neighbor algorithm, Central clustering

© 2023 Cuicui Wu, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

An analysis of the innovation strategy of student management in universities based on big data

Publié en ligne: 02 oct. 2023

Pages: -

Reçu: 18 oct. 2022

Accepté: 12 avr. 2023

DOI: https://doi.org/10.2478/amns.2023.2.00443

Mots clésStudent management, Behavior prediction, K-means clustering algorithm, K-nearest neighbor algorithm, Central clustering

© 2023 Cuicui Wu, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Mots clés
Student management, Behavior prediction, K-means clustering algorithm, K-nearest neighbor algorithm, Central clustering