본문 바로가기
  • Home

Prediction of Voice Phishing Victimization by Age Groups Using Principal Component Analysis: A Comparison of Regression Analysis and VAR Model Performance

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2025, 30(2), pp.33-43
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science
  • Received : January 8, 2025
  • Accepted : February 21, 2025
  • Published : February 28, 2025

Jungwoo Bae 1 Byunghong Choi 1

1경기과학고등학교

Accredited

ABSTRACT

Voice phishing victimization data exhibit distinct temporal patterns. This study compares the predictive performance of VAR and multiple linear regression models, considering their ability to capture time-series volatility. Principal component analysis was applied to voice phishing data (Jan 2018-Sep 2024), identifying three patterns explaining 99.31% of variation. An optimal window size of 12 months was determined through AIC testing to avoid overfitting. Model validation controlling for seasonal variability using Jul-Sep 2024 data showed the VAR model (MAE 33.06, MAPE 14.72%, RMSE 56.41, R² 0.94) outperformed multiple linear regression by 2-3 times across all metrics. The VAR model was then used to predict victimization for February 2025, with 95% confidence intervals accounting for data volatility. Results predicted decreased victimization across all age groups, leading to suggestions for age-specific prevention strategies.

Citation status

* References for papers published after 2023 are currently being built.