Computer Science ›› 2017, Vol. 44 ›› Issue (1): 243-246.doi: 10.11896/j.issn.1002-137X.2017.01.045

Previous Articles     Next Articles

Sparse Controllable Principal Component Analysis Method

TAN Ya-fang, LIU Juan, WANG Cai-hua and JIANG Wan-wei   

  • Online:2018-11-13 Published:2018-11-13

Abstract: Principal component analysis (PCA) is a multivariate statistical analysis method which chooses a few important variables (dimension reduction) by linear transformation.PCA is widely used in scientific researches and enginee-ring,however,the results can sometimes be difficult to interpret.Therefore,some researchers introduced sparse penalties (lasso,fused lasso and adaptive lasso etc.) to obtain interpretable results.Since the traditional sparse penalty is not easy to control,we presented a novel penalty,namely sparse controllable penalty (SCP),to control the sparsity of principal components.Compared with the traditional penalties,SCP is scale insensitive,dimension insensitive and bounded between 0 and 1.It is easy to adjust the super parameter to control sparseness.Experimental results demonstrate that sparse controllable principal component analysis (SCPCA) is efficient.

Key words: Principal component analysis,Sparse penalty,Sparse controllable principal component analysis

[1] JOLLIFFE I T.Principal Component Analysis(second ed.)[M].New York,Springer,2002.
[2] HASTIE T,TIBSHIRANI R,FRIEDMAN J.The Elements of Statistical Learning [M].Data mining,Interface and Prediction New York,Springer,2001.
[3] HANCOCK P J B,BURTON A M,BRUCE V.Face processing:human perception and principal components analysis [J].Memory and Cognition,1996,24(1):26-40.
[4] MISRA J,SCHMITT W,et al.Interactive Exploration of Microarray Gene Expression Patterns in a Reduced Dimensional Space [J].Genome Research,2012,12(7):1112-1120.
[5] SHEN Hai-peng,HUANG Jian-hua.Sparse principal component analysis via regularized low rank matrix approximation [J].Journal of Multivariate Analysis,2008,99(6):1015-1034.
[6] JOLLIFFE I T,UDDIN M.The Simplified Component Tech-nique:An Alternative to Rotated Principal Components[J].Journal of Computational and Graphical Statistics,2000,9(9):689-710.
[7] JOLLIFFE I T,TRENDAFILOV N T,et al.A Modified Principal Component Technique Based on the LASSO [J].Journal of Computational and Graphical Statistics,2003,12(3):531-547.
[8] ZOU H,HASTIE T,et al.Sparse principal component analysis [J].Journal of Computational and Graphical Statistics,2006,15:265-286.
[9] WITTEN D M,et al.A penalized matrix decomposition,withapplications to sparse principal components and canonical correlation [J].Biostatistics,2009,10(3):515-534.
[10] TIBSHIRANI R.Regression shrinkage and selection via the lasso [J].Journal of the Royal Statistical Society,1996,58(1):267-288.
[11] ZOU H.The adaptive lasso and its oracle properties [J].Journal of the American Statistical Association,2006,101(476):1418-1429.
[12] ALLEN G I,GROSENICK L,et al.the A Generalized Least-Square Matrix Decomposition [J].Journal of the American Statistical Association,2014,109(505):145-159.
[13] QI Xin,LUO Rui-yan,ZHAO Hong-yu.Sparse principal component analysis by choice of norm[J].Journal of Multivariate Analysis,2013,114(2):127-160.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!