A Supervised Correlation Coefficient Method: Detection of Different Correlation

Correlation coefficient is a kind of feature selection method that can effectively choose the most appropriate feature and reducing data dimensions. There are several supervised correlation coefficient methods have been considered. However, few methods can detect both the linear correlation and non-linear correlation. Our study considered a supervised correlation coefficient method, named consistency detection, to detect linear and non-linear correlation. Our results show consistency detection can better retain the dimensions of classification and detect correlation than Pearson correlation coefficient in some way.