drewlee
AOC, AUC, ROC这三个东西都是曲线吗?横纵坐标分别是什么?
false negative是不是指真实情况是H1,而没有拒绝H0?
false positive是不是指真实情况是H0,而拒绝了H0?
neige
I only know ROC, Y-axis is your sensitivity(1-false -), X axis is 1- specificity (false +)
In hypo testing, false + is your type I error, prob of rejecting H0 when it is true, in sensitivity and specificity analysis, it should be 1-specificity.
false - is your type II error, accept H0 when it is false, i.e. 1- power, so it will be
1-sensitivity
neige
a little update there
drewlee
Thanks a lot.
I supplement a paper about the interpretation of ROC and AUC ( Area Under the Curve ) :
http://www.cs.iastate.edu/~honavar/ROC101.pdf
neige
AUC, first time heard of it, is it just the C-statistics?
drewlee
[quote]引用第4楼neige于2007-05-28 12:15发表的“”:
AUC, first time heard of it, is it just the C-statistics?[/quote]
I am sorry that I don't know C-statistics, but I know that AUC is the area under curve of ROC.
[quote]The AUC has an important statistical property: the AUC of a
classifier is equivalent to the probability that the classifieer will rank
a randomly chosen positive instance higher than a randomly chosen
negative instance. This is equivalent to the Wilcoxon test of ranks
(Hanley and McNeil, 1982). The AUC is also closely related to the
Gini coefficient (Breiman et al., 1984), which is twice the area between
the diagonal and the ROC curve. Hand and Till (2001) point out that
Gini + 1 = 2 AUC.[/quote]
Reference: ROC Graphs: Notes and Practical Considerations for Researchers (2004)
neige
yeah, then it is the C-statistics, can be produced by WILCOXON AND MANN-WHITNEY TESTS
每天都不好
基本都是诊断试验中的词,AUC是指ROC曲线下的面积,ROC是人工受试着工作特征曲线. AOC没见过.
FP是非患者被诊断试验错误的诊断为阳性,也就是误诊率,相当于一型错误
FN是患者被错误诊断为阴性,漏诊率, 相当于二型错误
AUC可以用WILCOXON AND MANN-WHITNEY 统计量的非参数方法来估计