2,143 followers
Another bit of feedback: I strongly encourage you to use the evaluation framework by Pellegrino et al., so that results are comparable across papers and not influenced, e.g., by hyperparameters of the downstream classificiation or regression learner: https