本文介绍了[自动ML] [UI] - 错误的ROC图表的处理方法,对大家解决问题具有一定的参考价值,需要的朋友们下面随着小编来一起学习吧!

问题描述

我已经通过UI创建了一个AutoML(二进制分类)实验,可以得到一个相当准确的模型(准确度在86%左右)。 

I have created an AutoML (Binary Classification) experiment through the UI that results in a decently accurate model (accuracy around the 86% mark). 

生成的图表看起来很奇怪。 例如,ROC图表的曲线方式低于该线。  AUC很高,大约在90左右,所以我很困惑为什么我有这么糟糕的ROC图表。

The graphs produced seem very odd.  For example, the ROC Chart have the curve way below the line.  The AUC is high, around the 90 mark so I'm confused as to why I have such a bad looking ROC Chart.

我已经包含了ROC和混淆矩阵

I've included the ROC and the Confusion Matrix

http://blogs.adatis.co.uk/blogs/sachatomey

http://blogs.adatis.co.uk/blogs/sachatomey

推荐答案


这篇关于[自动ML] [UI] - 错误的ROC图表的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持!

11-03 14:45