Comparison Robust Biweight Midcovariance and Minimum Covariance Determinant Methods in Canonical Correlation Analysis

Abstract

Canonical Correlation Analysis (CCA) is a multivariate linear used to identify and quantify associations between two sets of random variables. Its standard computation is based on sample covariance matrices, which are however very sensitive to outlying observations. The robust methods are needed. There are two robust methods, i.e robust Biweight Midcovariance (BICOV) and Minimum Covariance Determinant (MCD) methods. The objective of this research is to compare the performance of both methods based on mean square error. The data simulations are generated from various conditions. The variation data consists of the proportion of outliers, and the kind of outliers: shift, scale, and radial outlier. The performance of robust BICOV method in CCA is the best compared to MCD and Classic.