search

amueller / introduction_to_ml_with_python

Notebooks and code for the book "Introduction to Machine Learning with Python"
7.45k stars 4.57k forks source link

about precision-recall curve #139

Open jingxigui opened 4 years ago

jingxigui commented 4 years ago

It seems that more often, the x-axis is recall and the y-axis is precision. https://scikit-learn.org/stable/auto_examples/model_selection/plot_precision_recall.html

amueller commented 4 years ago

Thanks! I agree that should be changed in the book. I'm actually surprised I did it this way around. The one benefit is then you have recall on the y axis in both the PR curve and ROC curve but indeed that's not the standard convention.