Research on Evaluation Methods for GPR-SEM Prediction Models 


Vol. 14,  No. 1, pp. 14-20, Jan.  2025
https://doi.org/10.3745/TKIPS.2025.14.1.14


PDF
  Abstract

The development of accurate prediction models has become essential in many aspects of manufacturing companies,such as quality control and product lifecycle management,due to advancements in machine learning technology. However, there are limitations in applying these models to fields where it is difficult to accumulate large amounts of high-quality data (e.g., new material development). Therefore, In previous research, we proposed the GPR-SEM model for small-dataset based prediction models. Traditional model fit indicators (such as MSE, R²) are useful for evaluating the performance of existing prediction models, but they have limitations in fully reflecting the uncertainty information, which is a characteristic of the GPR-SEM model. To address this, in this study, we propose the GS-Score, a comprehensive scoring method for evaluating the GPR-SEM model. By integrating various evaluation metrics into a single score that represents the overall performance of the model, it facilitates comparison and selection between models. In this study, we applied traditional evaluation methods to small-datasets and verified their effectiveness through comparison and contrast with the GS-Score. Through this research, we expect to establish objective evaluation criteria for developing and applying prediction models using GPR-SEM.

  Statistics
  Cite this article

[IEEE Style]

M. Kyung-Yeol and P. Kun-Uk, "Research on Evaluation Methods for GPR-SEM Prediction Models," The Transactions of the Korea Information Processing Society, vol. 14, no. 1, pp. 14-20, 2025. DOI: https://doi.org/10.3745/TKIPS.2025.14.1.14.

[ACM Style]

Moon Kyung-Yeol and Park Kun-Uk. 2025. Research on Evaluation Methods for GPR-SEM Prediction Models. The Transactions of the Korea Information Processing Society, 14, 1, (2025), 14-20. DOI: https://doi.org/10.3745/TKIPS.2025.14.1.14.

Forms

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)

Old Journals

Indexing

All publications of TKIPS is indexed in DOI, EBSCO, Google Scholar, Crossref, and CrossCheck.

TKIPS is also selected as the Journal for Accreditation by NRF (National Research Foundation of Korea).

Related Journals