Generating Multivariate Exam Scores using Copulas and Socioeconomic Factors

Jackson Tsz Wah Chan, Kwok Tai Chui, Leung Pun Wong, Ryan Wen Liu, Kwan Keung Ng, Yan Keung Hui

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Final exams are essential to assessing students' level of understanding of the course materials and evaluating their achievements in the learning outcomes. Many factors may have affected their performance throughout the courses. These are personal-, social-, and economic-related. Student Exam Performance Datasets are usually small-scale. The nature of a typical dataset comprises a small class size and few assessment components, and the students and parents may refuse to disclose and share too much personal information. This paper proposes a copula-based data generation algorithm that provides additional training data for the datasets. The algorithm is evaluated based on eight aspects: diagnostic, data quality, missing value similarity, statistic similarity, category coverage, range coverage, new row synthesis, and column comparison.

Original languageEnglish
Title of host publicationProceedings - 2024 International Symposium on Educational Technology, ISET 2024
EditorsKwok Tai Chui, Yan Keung Hui, Dingqi Yang, Lap-Kei Lee, Leung-Pun Wong, Barry Lee Reynolds
Pages75-79
Number of pages5
ISBN (Electronic)9798350361414
DOIs
Publication statusPublished - 2024
Event10th International Symposium on Educational Technology, ISET 2024 - Macao, China
Duration: 29 Jul 20241 Aug 2024

Publication series

NameProceedings - 2024 International Symposium on Educational Technology, ISET 2024

Conference

Conference10th International Symposium on Educational Technology, ISET 2024
Country/TerritoryChina
CityMacao
Period29/07/241/08/24

Keywords

  • Copula
  • data generation
  • exam score
  • socioeconomic factors
  • student performance

Fingerprint

Dive into the research topics of 'Generating Multivariate Exam Scores using Copulas and Socioeconomic Factors'. Together they form a unique fingerprint.

Cite this