I discovered that the problem was a conflict between the outdated NaN package (which includes subfunctions titled nanmean.m and mean.m) and the preinstalled mean.m. Deleting the NaN package and using mean(x,'omitnan') now solves the issue.
Error 'Matrix dimensions must agree' when using nanmean
3 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Colin Raymond
am 20 Jun. 2020
Beantwortet: Colin Raymond
am 20 Jun. 2020
I'm on Mac 10.14.6 and Matlab R2019a. Nanmean and mean have always worked as expected, but suddenly they have begun throwing 'matrix dimensions must agree' errors for any and all inputs (for mean, this is only when using the 'omitnan' option). This happens even on simple 1D arrays, such as
nanmean([5;2;6;NaN;7])
The error is:
Matrix dimensions must agree.
Error in mean (line 105)
elseif any(opt=='aAgGhH')
Error in nanmean (line 22)
y = mean(varargin{:},'omitnan');
I haven't updated Matlab recently, or changed the environment (at least to my knowledge). Any idea what could be going on?
Interestingly
sum([5;2;6;NaN;7],'omitnan')/5
returns the expected result of 4, however of course this is not nearly as elegant as nanmean().
0 Kommentare
Akzeptierte Antwort
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Resizing and Reshaping Matrices finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!