See all my videos at www.tilestats.... 1. Example data (FP and TP) 2. How to calculate the FDR (06:12) 3. FDR vs FWER (07:26) 4. How to interpret the FDR (09:35)
Thank You very much for the clear and great explanation! But I cannot get the following - the formula of FDR is given as ration of FP to sum of FP and TP. But on the minute 7.15 of the video there is also E in the calculation... How is that used, why is it there?
I ran a simulation based on the example with 5000 true H0 and 5000 false H0. Due to chance, I rejected 270 true H0 (270 out of the 5000 p-values happened to be less than 0.05). Since we use a significance level of 5%, we expected to reject 250 true H0, but due to chance, in the simulations, I rejected instead 270 true H0. 3288 TP is also a result of the same simulation.
Excellent explanation! Thank you :)
Thank you!
Underrated
EXCELLENT..HEARTFELT THANKS ...
Thank You very much for the clear and great explanation! But I cannot get the following - the formula of FDR is given as ration of FP to sum of FP and TP. But on the minute 7.15 of the video there is also E in the calculation... How is that used, why is it there?
Don’t worry about E, it is just a notation for ”expected”. It is not included in the calculations.
@@tilestats thanks a lot!
Are 270 and 3288 arbitrary numbers you chose? Many thanks for the video.
I ran a simulation based on the example with 5000 true H0 and 5000 false H0. Due to chance, I rejected 270 true H0 (270 out of the 5000 p-values happened to be less than 0.05). Since we use a significance level of 5%, we expected to reject 250 true H0, but due to chance, in the simulations, I rejected instead 270 true H0.
3288 TP is also a result of the same simulation.
At 4:16, why would the distribution be uniform when null hypothesis is true.
Yes, that is not obvious so I made this video:
ua-cam.com/video/aYqIs4XZli8/v-deo.html
TG