# nanmean different outcomes?

2 views (last 30 days)
JamJan on 15 Mar 2019
Commented: JamJan on 15 Mar 2019
Hi,
I'm doing some analysis on data that has the following time intervals, however when I calculate the mean on two different ways (just for sanity) I ran into the following problem.
I have two arrays:
A = [NaN 1,57031250000000 1,60156250000000 1,71093750000000 1,65234375000000 1,51367187500000 1,49804687500000 1,49414062500000 1,59179687500000 1,41015625000000];
nanmean(A) = 1.5603
B = [NaN 1,64648437500000 0,796875000000000 0,667968750000000 1,65820312500000 1,69140625000000 1,39062500000000 1,66015625000000 1,40039062500000 1,57226562500000 1,65429687500000];
nanmean(B) = 1.4139
Now I have put them behind each other in a [A B] way.
C = [A B];
nanmean(C) = 1.4832
However this mean is not the same as the mean of A and B done separately.
(1.5603 + 1.4139)/2 = 1.4871
How is this possible and why is it different? Is this because of the NaNs?

Alex Mcaulley on 15 Mar 2019
Your two arrays don't have the same number of non nan elements (10 vs 11), then the mean is
(1.5603*sum(~isnan(A)) + 1.4139*sum(~isnan(B)))/(sum(~isnan(A))+sum(~isnan(B))) = 1.4832
JamJan on 15 Mar 2019
Thank you, indeed very logical!