Need for nested cross-validation?
A general question for discussion here: Currently, Aryan performs a non-nested cross-validation. For the random forest, are there any hyperparameters that are set manually or that are chosen even data-dependently? If yes, overall performance may be biased, and a nested cross-validation scheme that includes the hyperparameters might be a good idea:
See https://scikit-learn.org/stable/auto_examples/model_selection/plot_nested_cross_validation_iris.html, for example