参考
- https://www.w3cschool.cn/doc_scikit_learn/scikit_learn-auto_examples-model_selection-plot_precision_recall.html?lang=en#sphx-glr-auto-examples-model-selection-plot-precision-recall-py
- https://mercurixito.github.io/2021/05/22/metric-map/
- https://blog.csdn.net/mr_muli/article/details/101616406
- skearn包的阈值是卡阈值，所以相同阈值取后面，
- https://zhuanlan.zhihu.com/p/33372046
- https://www.cnblogs.com/ywheunji/p/13376090.html
- https://scikit-learn.org.cn/view/496.html
- https://blog.51cto.com/liguodong/4925397
3d图像绘制 - https://www.geogebra.org/3d?lang=zh_CN
map计算 - https://zhuanlan.zhihu.com/p/33372046
map - https://mercurixito.github.io/2021/05/22/metric-map/
有点意思 - https://zdkswd.github.io/categories/
roc曲线是
TPR召回 和 FPR - https://zdkswd.github.io/2018/11/20/精确率、召回率、真正类率（TPR）、假正类率（FPR）ROC曲线/
绘制roc - https://blog.csdn.net/hesongzefairy/article/details/104302499
- https://blog.csdn.net/pearl8899/article/details/110410827
