How to replace values in a cell array based on a condition?

I have a cell array 'C' that is 1x525, with each cell holding a matrix with 18 columns and varying rows. I also have cell arrays 'F' and 'G', which are also 1x525, and contain matrices of size 1x18. What I want is: if element C{1,1}(10,14) is greater than F{1,1}(1,14), the element in C should be replaced with said value in F. If C{1,1}(10,14) is less than G{1,1}(1,14), it should be replaced with said value in G. I want to do this for every cell in C with its corresponding cell in F and G, with respect to only columns 8:18. Any ideas are welcome. Here is my code so far as to how I arrived at C, F, and G:

load weekdata.mat
N = size(weekdata,3);
C = cell(1,N);    % weekly data
D = cell(1,N);    % means
E = cell(1,N);    % standard deviations
F = cell(1,N);    % 3 z-scores above
G = cell(1,N);    % 3 z-scores below
for k = 1:N
    idx = weekdata(:,3,k)==0 & weekdata(:,4,k)==0 & weekdata(:,5,k)==1 & weekdata(:,6,k)==0;
    tmp = weekdata(idx,:,k);
    C{k} = tmp;
    if  size(C{1,k},1) > 1
        D{1,k} = mean(C{1,k});
        E{1,k} = std(C{1,k});
    else
        D{1,k} = C{1,k};
        E{1,k} = zeros(1,18);
    end
    F{k} = D{k}+3*E{k};
    G{k} = D{k}-3*E{k};
end

Answer 1

[Stephen23]

for k = 1:numel(C)
    C{k}(:,8:18) = min(F{k}(:,8:18),max(G{k}(:,8:18),C{k}(:,8:18)));
end

[Neal Mannan Hasan]

Have tried plugging this in, as follows:

N = size(weekdata,3);
C = cell(1,N);    % weekly data
D = cell(1,N);    % means
E = cell(1,N);    % standard deviations
F = cell(1,N);    % 3 z-scores above
G = cell(1,N);    % 3 z-scores below
A = cell(1,N);
for k = 1:N
    idx = weekdata(:,3,k)==0 & weekdata(:,4,k)==0 & weekdata(:,5,k)==1 & weekdata(:,6,k)==0;
    tmp = weekdata(idx,:,k);
    C{k} = tmp;
    if  size(C{1,k},1) > 1
        D{1,k} = mean(C{1,k});
        E{1,k} = std(C{1,k});
    else
        D{1,k} = C{1,k};
        E{1,k} = zeros(1,18);
    end
    F{k} = D{k}+3*E{k};
    G{k} = D{k}-3*E{k};
end
for k = 1:numel(C)
    C{k}(:,8:18) = min(F{k}(:,8:18),max(G{k}(:,8:18),C{k}(:,8:18)));
end

But I get the following error:

Error using max
Matrix dimensions must agree.

Error in multi_factor_analysis (line 25)
    C{k}(:,8:18) = min(F{k}(:,8:18),max(G{k}(:,8:18),C{k}(:,8:18)));

Any ideas?

[Stephen23]

@Neal Mannan Hasan: what MATLAB version are you using? For versions pre R2016b you might need to use bsxfun for each of min and max, e.g.:

for k = 1:numel(C)
    tmp = C{k}(:,8:18);
    tmp = bsxfun(@max,tmp,G{k}(:,8:18));
    tmp = bsxfun(@min,tmp,F{k}(:,8:18));
    C{k}(:,8:18) = tmp;
end

[Neal Mannan Hasan]

Ah that fixed it, I'm on R2014a, this company runs a pretty old version :) However, I want to replace values greater than F and less than G; does this do that or the opposite? I only ask as the @max is with G{k} and @min with F{k}, but perhaps I misunderstand the functionality?

[Stephen23]

"I only ask as the @max is with G{k} and @min with F{k}, but perhaps I misunderstand the functionality?"

Your question states "If C{1,1}(10,14) is less than G{1,1}(1,14), it should be replaced with said value in G".

This means that you want the output to have the maximum of the values in C and G: if C is less than G and you replace C with G then this is the maximum of C and G. You can easily check this yourself, and decide if it is doing what you want it to do:

>> C = [2,4,8;1,2,3]
C =
   2   4   8
   1   2   3
>> G = [0,3,9];
>> bsxfun(@max,C,G)
ans =
   2   4   9
   1   3   9

[Neal Mannan Hasan]

I see. Thanks Stephen!