FPR should really be "1-specificity" not "specificity" 😅

in #think4 years ago

Bit of a long shot but is anyone familiar with ROC curves? I'm trying to calculate the area under the curve (AUC), which in theory should be simple enough as I have already calculated the sensitivity and specificity values for various cut-off points and I've even plotted a curve, but without statistical software I'm stumped. Apparently it takes only 2mins to calculate it using R or SPSS if you know how! I thought I nearly cracked it using a very basic version of Excel but it doesn't look right. I've been in the hospital library for nearly 8 hours and my brain is fried! 😭🙏

Below is my ROC curve and the data I based it on which I'm certain I've done correctly 😅 and my attempt on Excel at calculating the AOC which is probably utter nonsense the highlighted parts are what I'm unsure about.

p.jpg

p1.jpg

p2.jpg

Not that I think anyone will be reading this (I guess you never know, quite a few of you commented on my last stats SOS post!) but this is another table I did earlier which is more accurate in that it fits exactly with my ROC curve but FPR should really be "1-specificity" not "specificity" as I've done here in the table even though it confusingly seems to come out accurately as 1-specificity in the ROC curve.

pp.jpg

Coin Marketplace

STEEM 0.30
TRX 0.12
JST 0.036
BTC 66992.58
ETH 3460.53
USDT 1.00
SBD 4.68