I have this PR AUC plot, with both PCA and autoencoder related curves having a huge drop of precision in the beginning and then increasing again, with PCA hitting 0 as you can see in the zoomed in image.
Is this correct? What could be the cause for it? I can't seem to find an explanation for this different behavior from other PR AUC curves.
(the dataset is heavily imbalanced, only having 0,22% of positive class instances, I suspect this has something to do with it)
(There is a similar discussion here AUC-PR but there is no recall or precision, but I couldn't understand the answer and I just created an account so I can't comment yet...)
