The coefficient of discrimination, denoted 𝑅², is a commonly used performance metric for regression. It provides a measure of the proportion of the variance of a dependent variable that is explained by a regression model and defined by
Where 𝑦𝑖 is a dependent variable, 𝑦̂𝑖 is the output of the regression model, both indexed by 𝑖, and ӯ is the mean of the dependent variable. 𝑅² is always less than or equal to 1; the larger it is, the more the variance of the dependent variable is explained by the regression model.
The coefficient of discrimination is important because it is a commonly used and accepted performance metric for regression models.
How C3 AI Helps Organizations Use Coefficient of Discrimination, R-Squared (R2)
The coefficient of discrimination, or 𝑅², can be used readily as a performance metric by setting the scoring metric of a C3 MLPipe to MLRSquaredMetric.