I have a taste for graphs with ln-odds as the x-axis over probability… but I'm having a hard time forming a supporting argument. any help?
-
-
Replying to @snoble
it has something to do with going from 95% of success to 97.5% is a big deal (halves the chance of failure)
1 reply 0 retweets 0 likes -
Replying to @avibryant
@avibryant you get both for free with ln-odds (but I don't find this reason all that compelling tbh)1 reply 0 retweets 0 likes -
Replying to @snoble
@avibryant I think probability is what you look at when you care about aggregate outcomes (what's the EV)…1 reply 0 retweets 0 likes -
Replying to @snoble
@avibryant …ln-odds is what you care about when making decisions for cut offs for individuals (what's the cost to halve failure?)1 reply 0 retweets 0 likes
Replying to @snoble
@snoble I feel like what you might really want here is to just plot a loss function that takes cost of type I/II errors into account, then.
12:34 PM - 22 Nov 2014
from Richmond, British Columbia
0 replies
0 retweets
0 likes
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.