The main limitation of correlation coefficient is that it cannot be applied to non-linear model as it only measures the degree of linear association between two variables, X and Y. Nevertheless, some statistics practitioners are not aware of this fact and use it in non-linear model. Consequently, misleading conclusions are obtained from using incorrect measures. In this project we propose restructuring the Mutual Information (MI) measure so that it is able to measure non-linear dependency between variables. The new MI provides easier interpretation of the dependence between two variables in non-linear model
