Improved Estimation Of Generalized Measures Of Correlation

The relationship of random variables has played an important role in statistics, Pear-son correlation coefficient has been widely used as a measure of explained variance, but ithas some limitations, for example, it only deals with linear relations between random vari-ables. In order to develop broad applicable correlation measures, Zheng et al.(2012)[1]de-velop a pair of generalized measures of correlation which deal with asymmetries in ex-plained variances, and nonlinear relations between random variables. Zheng et al.(2012)[1]use kernel method to estimate GMC, but they do not account for the mutuality in band-width matrix, this paper will account for it, use kernel method to estimate GMC again, andcompare with the results of Zheng et al.(2012)[1]. Simulation results will show advantagesand shortcomings in estimating GMC.