There is an existing function in sklearn for calculating accuracy for TopK labels predicted in classification task (top_k_accuracy_score()). Are there any analogues for calculating other evaluation metrics, like Precision-Recall for TopK labels? So the idea that we take predict_proba() output as a base and True Positive(TP) for us will be situation that if top 3 of probabilities prediction for this observation will contain probability for that label.
Many thanks.
