Selection of the Number of Components to Keep in Order to Build High Quality Regression Model

Solovei Olga, Solovei Bohdan

Abstract: In the article is considered a question of how to identify the number of components to keep when using principal components analysis technique for dimension reduction. In paper is presented the impact of incorrectly chosen number on the quality of regression model; reasons why the number can be identified incorrectly using principal components analysis. The summary proposes the method for identification the number of components to keep which can be used together with principal components analysis technique for dimension reduction.

Keywords: Principal component analysis PCA, Singular Value Decomposition SVD, eigenvalues, Kaiser rule