Critical analysis of the coefficient of determination (R2), as an indicator of the quality of linear and non-linear models

Abstract

In almost all areas of science, data modeling has been used over the years for different purposes, mainly explanatory and predictive. To test whether the applied models fit the data, the coefficient of determination (R2) is often used. This article propose the pros and cons of its use as an indicator of the quality of linear and non-linear models and the possible solutions to problems arising from its application. The need to test the parameters of the model and the applicability of the number of parameters to be considered within the model used is also analyzed, an aspect that influences the quality of the fit.