응용통계연구 (The Korean Journal of Applied Statistics)

  • 제36권4호
  • /
  • Pages.279-294
  • /
  • 2023
  • /
  • 1225-066X(pISSN)
  • /
  • 2383-5818(eISSN)
한국통계학회 (The Korean Statistical Society)
권호 표지
DOI QR코드

DOI QR Code

주성분 분석을 활용한 재현자료 생성

Synthetic data generation by probabilistic PCA

  • 투고 : 2023.01.02
  • 심사 : 2023.04.17
  • 발행 : 2023.08.31
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

재현자료를 생성할 때 순차회귀 다중대체(SRMI)를 이용하는 방식이 가장 널리 알려져 있으며, 이를 구현한 소프트웨어로 R-패키지 synthpop이 활용되고 있다. 본 논문에서는 확률적 주성분 분석(PPCA)을 이용하여 재현자료를 생성하는 방안을 제안하고 2개의 데이터 세트를 이용한 모의실험으로 SRMI 방식과 PPCA 방식을 비교하였다. 모의실험에서 PPCA 방식으로 생성한 재현자료는 쌍별 상관계수를 기준으로 원자료와의 유사성이 가장 우수함을 확인하였다. 향후 PPCA 방식을 이용하여 시계열 자료에 대한 재현자료 생성을 연구하고자 한다.

It is well known to generate synthetic data sets by the sequential regression multiple imputation (SRMI) method. The R-package synthpop are widely used for generating synthetic data by the SRMI approaches. In this paper, I suggest generating synthetic data based on the probabilistic principal component analysis (PPCA) method. Two simple data sets are used for a simulation study to compare the SRMI and PPCA approaches. Simulation results demonstrate that pairwise coefficients in synthetic data sets by PPCA can be closer to original ones than by SRMI. Furthermore, for the various data types that PPCA applications are well established, such as time series data, the PPCA approach can be extended to generate synthetic data sets.

키워드

참고문헌

  1. Abdi H and Williams LJ (2010). Principal component analysis, Computational Statistics, 2, 433-459. https://doi.org/10.1002/wics.101
  2. Domingo-Ferrer J, Muralidhar K, and Bras-Amoros M (2021). General confidentiality and utility metrics for privacy-preserving data publishing based on the permutation model, IEEE Transactions on Dependable and Secure Computing, 18, 2506-2517. https://doi.org/10.1109/TDSC.2020.2968027
  3. Jolliffe IT (1986). Principal Component Analysis, Springer, New York.
  4. Jolliffe IT and Cadima J (2016). Principal component analysis: A review and recent developments, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374, 20150202.
  5. Nowok B, Raab GM, and Dibben C (2016). synthpop: Bespoke creation of Synthetic data in R, Journal of Statistical Software, 74, 1-26. https://doi.org/10.18637/jss.v074.i11
  6. Kim J and Park M-J (2019). Multiple imputation and synthetic data, The Korean Journal of Applied Statistics (written in Korean), 32, 83-97. https://doi.org/10.5351/KJAS.2019.32.1.083
  7. Kim HJ, Reiter JP, and Karr AF (2018). Simultaneous edit-imputation and disclosure limitation for business establishment data, Journal of Applied Statistics, 45, 63-82. https://doi.org/10.1080/02664763.2016.1267123
  8. Park M-J, Han J, and Park N (2020). Study on Synthetic Data Generation Methods with Applications to Statistics Korea RDC Data (written in Korean), Statistics Research Institute, Statistics Korea.
  9. Tipping ME and Bishop CM (1999). Probabilistic principal component analysis, Journal of the Royal Statistical Society Series B: Statistical Methodology, 61, 611-622. https://doi.org/10.1111/1467-9868.00196
  10. Roweis S (1998). EM algorithms for PCA and SPCA. In M Kearns, M Jordan, and S Solla (Eds), Advances in Neural Information Processing Systems 10: Proceedings of the 1997 Conference (pp. 626-632), The MIT Press.
  11. Stacklies W, Redestig H, Scholz M, Walther D, and Selbig J (2007) pcaMethods - a bioconductor package providing PCA methods for incomplete data, Bioinformatics, 23, 1164-1167. https://doi.org/10.1093/bioinformatics/btm069
  12. Wasserman L (2010). All of Statistics, Springer, New York.