Hacker News Reader

I never know how to interpret accuracy ratings for tests. If I told everyone I ran into on the street that they did not have Lou Gehrig's disease, regardless of any testing, I would still be 99.993% accurate. Can someone shed some insight on what 86% means?

View on HN

Yeah, I think it be comes intuitive as soon as you break it into % false positive and % false negative. Then, after you have test results in hand, you know your odds. In order to use the information from the test, you have to branch on your conclusion given the result of the test.

edit: then I read the link, and apparently both numbers are basically coincidentally the same here.

Reply on news.ycombinator.com