I'm using the Gini coefficient to evaluate the performance of a model. Making some changes (feature selection, hyperparameter tuning, etc.) I created variant models with different Gini coefficients.
How can I prove that the improvement in the Gini coefficient is indeed statistically significant?
In R there is a package called DescTools within it you can find a method for calculating Gini Coefficients, where you can set a confidence level, and get intervals back using a bootstrap method (simulation)
This is how I have seen it done, and probably as meaningful as you need in most situations because you can set alpha to be as conservative as you need.
