When designing a neural network, are there architectures that can be used to favor precision vs. recall and vice versa?
Contents
hide
Best Answer
The best way is to use ROC curve to find the optimal point, you can change the output threshold and get TP Rate , FP rate and plot ROC curve.
From ROC curve, you can get the optimal threshold value, to have the result you want.
Similar Posts:
- Solved – What are the differences between AUC and F1-score
- Solved – Average multiple precision recall curves
- Solved – Precision – Recall: Graphical Representation
- Solved – Trying to understand reasons behind low true positive rate in confusion matrix
- Solved – The representation of F1-score on the Precision-Recall Curve