MATLAB: Subscript indices must either be real positive integers or logicals in bsxfun

bxsfunsubscript indices

Hi there,
Had a look at various others posts that include this error message but cannot find what is missing from my code. Feel like I am brute forcing this by as it stands I have
T     19081 x 22 double % Raw Data values 
StdTimedEdit  1 x 22 double % Standard Deviation values for each column
MeanTimedEdit 1 x 22 double % Means for each column
StDIndexFull  19081 x 22 logical % Logical Index for which values I want to run the bxs function on
I am trying to do run a bxsfun to replace values outside 2 standard deviations from the mean for each column with 'NaN', but only on the values that equal 1 in the logical StDIndexFull. Currently it looks as such:
T(bsxfun(@lt, T(T(StdIndexFull)), (MeanTimedEdit + 2*StDsTimedEdit)) ...
& bsxfun(@gt, T(T(StdIndexFull)), (MeanTimedEdit - 2*StDsTimedEdit)))= NaN;
This returns the following error:
Subscript indices must either be real positive integers or logicals.
Any assistance or opinions would be muchly appreciated
Matt

Best Answer

As Star said, one of your problem is the use of T to index into itself. That makes no sense, I'm not sure what you intended with that.
In addition, your logic has gone a bit wrong on where to use the logical array. You want to find the values that:
  1. are outside the standard deviation (using bsxfun indeed) AND
  2. correspond to true in the logical array
This is written as:
(bsxfun(@lt, T, MeanTimedEdit - 2*StDsTimedEdit) | bsxfun(@gt, T, MeanTimedEdit + 2*StDsTimedEdit)) ... %part 1
& StdIndexFull  %part 2
So the replacing is written as:
T((bsxfun(@lt, T, MeanTimedEdit - 2*StDsTimedEdit) | bsxfun(@gt, T, MeanTimedEdit + 2*StDsTimedEdit)) & StdIndexFull) = nan;
Related Question