Application of ICA for Multiple Regression

• 발행기관: 한국자료분석학회
• 저자명: 김대학(Daehak Kim)
• 간행물 정보: 『Journal of The Korean Data Analysis Society (JKDAS)』Vol.7 No.3, 721~728쪽, 전체 8쪽
• 주제분류: 자연과학 > 통계학
• 파일형태: PDF
• 발행일자: 2005.06.30

영문 초록

Independent Component Analysis(ICA) is a widely applicable modern method for finding underlying factors or components from multivariate statistical data. In this paper, we consider the application of ICA for multiple regression. When multicollinearity is present in a set of explanatory variables, the ordinary least squares estimate of the individual regression coefficient tend to be unstable and can lead to erroneous inferences. Numerical study shows we can have stable parameter estimates based on ICA regression when multicollinearity exist.

목차

1. Introduction

2. ICA

3. ICA Regression

4. Numerical Study

5. Concluding Remarks

References

키워드

Independent component analysis Principal component analysis Multiple regression Multicollinearity Independent component regression