Scaled PCA: A New Approach to Dimension Reduction

38 Pages Posted: 14 May 2019 Last revised: 27 Jan 2021

See all articles by Dashan Huang

Dashan Huang

Singapore Management University - Lee Kong Chian School of Business

Fuwei Jiang

Central University of Finance and Economics (CUFE)

Kunpeng Li

Capital University of Economics and Business

Guoshi Tong

Renmin University

Guofu Zhou

Washington University in St. Louis - John M. Olin Business School

Date Written: March 23, 2019

Abstract

This paper proposes a novel supervised learning technique for forecasting -- scaled principal component analysis (sPCA). The sPCA improves the traditional principal component analysis (PCA) by scaling each predictor with its predictive slope on the target to be forecasted. Unlike the PCA that maximizes the common variation of the predictors, the sPCA assigns more weights to those predictors with stronger forecasting power. In a general factor framework, we show that, under some appropriate conditions on data, the sPCA forecast beats the PCA forecast; and if these conditions break down, extensive simulations indicate that the sPCA still has a large chance to outperform the PCA. A real data example on macroeconomic forecasting shows that the sPCA has better performance in general.

Keywords: Forecasting, PCA, Big Data, Machine Learning, Supervised Learning

JEL Classification: C22, C23, C53

Suggested Citation

Huang, Dashan and Jiang, Fuwei and Li, Kunpeng and Tong, Guoshi and Zhou, Guofu, Scaled PCA: A New Approach to Dimension Reduction (March 23, 2019). Available at SSRN: https://ssrn.com/abstract=3358911 or http://dx.doi.org/10.2139/ssrn.3358911

Dashan Huang

Singapore Management University - Lee Kong Chian School of Business ( email )

50 Stamford Road
Singapore, 178899
Singapore

HOME PAGE: http://dashanhuang.weebly.com/

Fuwei Jiang

Central University of Finance and Economics (CUFE) ( email )

39 South College Road
Haidian District
Beijing, Beijing 100081
China

Kunpeng Li

Capital University of Economics and Business ( email )

Zhangjialukou 121, Huaxiang
Fengtai district
Beijing, 100070
China

Guoshi Tong

Renmin University ( email )

59 Zhongguancun Street
Beijing, 100872
China

Guofu Zhou (Contact Author)

Washington University in St. Louis - John M. Olin Business School ( email )

Washington University
Campus Box 1133
St. Louis, MO 63130-4899
United States
314-935-6384 (Phone)
314-658-6359 (Fax)

HOME PAGE: http://apps.olin.wustl.edu/faculty/zhou/

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
1,702
Abstract Views
6,371
Rank
19,190
PlumX Metrics