WSEAS Transactions on Computers
Print ISSN: 1109-2750, E-ISSN: 2224-2872
Volume 18, 2019
A Comparison of Monotonic Correlation Measures with Outliers
Authors: ,
Abstract: "This paper aims at examining the performance of a recently proposed measure of dependence – the Monotonic Dependence Coefficient – MDC - with respect to classical monotonic correlation measures like Pearson’s r, Spearman’s ?, and Kendall’s τ, using simulated outlier contaminated and non-contaminated data sets as well as a contaminated real dataset, considering three different cases. This comparison aims at checking how and when these coefficients detect dependence relationships between two variables when outliers are present. Several scenarios are created with multiple values for the dependence measures, outlier contamination fractions and data patterns. The basic simulated dataset is generated from a bivariate standard normal distribution. Using values generated from the exponential, power-transformed, lognormal, and Weibull distributions, added to the basic generated dataset, we transform the contaminated data, allowing for multiple patterns. The main findings tend to favour the Spearman’s ? coefficient for most of the simulated scenarios, especially when the outlier contamination is taken into account, whereas MDC performs better than ? in noncontaminated data. However, in the real data scenario Spearman’s ? outperforms the other measures in two out of three cases, whereas MDC performs better in the other case"
Search Articles
Keywords: "Outliers, Correlation Coefficient, Monotonic Dependence, Monte Carlo Simulation, Environmental Quality, Economic Growth"
Pages: 223-230
WSEAS Transactions on Computers, ISSN / E-ISSN: 1109-2750 / 2224-2872, Volume 18, 2019, Art. #29