It's really amazing!! Why are the calculation results of the AUC (recall, precision) function and the average precision score function significantly different?

[kourenke]

Method 1:
Precision, recall, _=Precision-Recall_curve()
A=auc (recall, precision)
Method 2:
B=average precision score ()
Method 1 and Method 2 both calculate the area under the PR curve, but why are the results significantly different, i.e. a ≠ b

[xuefeng-xu]

Quote from the document page.

average_precision_score is different from computing the area under the precision-recall curve with the trapezoidal rule.