mean of columns in a cell

I have a cell, wich looks like

'S001' [] ' M' 29 '70' '180' '118' '70'

'S002' [] 'F' 25 '58' '166' '130' '90'

'S003' [] 'M' 34 '62' '167' '120' '78'

'S004' [] 'F' 28 '72' '160' '118' '78'

...

Now I would like to have the means of each row. I try

cellfun(@mean,cell_formatiert(3:14,k))

I get an answer, but it is wrong.

[Franziska Domeier]

I think the problem is, that the values are scalars and not numbers, cause with column 4 it works.

[James Tursa]

What would be the desired answer for your example?

[Franziska Domeier]

If I do it with an index, the answer is

Conversion to cell from double is not possible.

If I do it for the columns I need, it is for the fourth:

ans =

    29
    25
    34
    28
    30
    34
    24
    27
    29
    27
    33
    38
    

and then I can make the mean of it. This works, bit wirth column 5 the answer is:

    ans =

   51.5000
   54.5000
   52.0000
   52.5000
   52.0000
   54.0000
   54.5000
   51.0000
   55.0000
   53.5000
   54.5000
   53.0000

and this doesn't make sense

[Adam Danz]

I'm confused. The title of the question states that you want the mean of columns. But then in your question, " I would like to have the means of each row". So, it is rows or columns?

[Franziska Domeier]

sorry, I mean column

Answer 1

[Adam Danz]

Inputs:

• cell_formatiert: your mxn cell array (presumably containing a mixture of numeric and character values, otherwise there is a much simpler answer).
• rows : a vector of row numbers to include in the analysis
• cols : a vector of column numbers to include in the analysis

cell_formatiert=  {
    'S001'	[]	'M'	29	'70'	'180'	'118'	'70'
    'S002'	[]	'F'	25	'58'	'166'	'130'	'90'
    'S003'	[]	'M'	34	'62'	'167'	'120'	'78'
    'S004'	[]	'F'	28	'72'	'160'	'118'	'78'};
% Select the columns to average for each row
cols = 4:8;
% Select the rows that should be analyzed
rows = 1:4;

Here we convert any character arrays that are within the cols and rows selection to numeric

% identify elements of the selected columns that are character arrays
charIdx = cellfun(@ischar,cell_formatiert);
charIdx(:,~ismember(1:size(cell_formatiert,2),cols)) = false; 
% Convert char arrays from chosen columns to numeric
cell_formatiert(charIdx) = num2cell(str2double(cell_formatiert(charIdx)));

If you want to comput the mean of each selected row, given the selected columns

% compute means per selected rows
mu = arrayfun(@(r)mean([cell_formatiert{r,cols}]),rows).'; 

mu is a vector of row-means, the same size as rows.

If you want to comput the mean of each selected column, given the selected rows

% compute means per selected columns
mu = arrayfun(@(c)mean([cell_formatiert{rows,c}]),cols); 

mu is a vector of column-means, the same size as cols.

[Franziska Domeier]

Sorry, I make an mistake in my question. First of all, thanks, for the row it works, but I need it for each column.

Sorry, can you help again please?!

[Adam Danz]

Check out the two bold lines of text in my answer. You'll want to use the last line.

[Franziska Domeier]

thanks a lot, it works !!!

[Adam Danz]

Glad I could help!