Double conversion to use prctile function

Instead of evaluating column by column . I include a for loop so I can use the cell2mat function to convert from double to whatever data type the prctile function uses. However, after the for loop, I get an error message:

"Error using double

Conversion to double from cell is not possible.

Error in prctile (line 55)

x = double(x);

Error in E99 (line 45)

P = prctile(Model_mat(1,:),(99))"

But if I individually convert the column with the cell2mat function- the percentile function works fine.

% This is column by column
Model_mat1 = cell2mat(Model(1,:));
Model_mat2 = cell2mat(Model(2,:));
Model_mat3 = cell2mat(Model(3,:));
Model_mat4 = cell2mat(Model(4,:));
Model_mat5 = cell2mat(Model(5,:));
Model_mat6 = cell2mat(Model(6,:));
Model_mat7 = cell2mat(Model(7,:));
Model_mat8 = cell2mat(Model(8,:));
Model_mat9 = cell2mat(Model(9,:));
Model_mat10 = cell2mat(Model(10,:));

% This is the for loop 
for i = 1:length(Model)
    %Model_mat{i,:} = cell2mat(Model(i,:)); 
    %Model_mat{i,:} = int64(Model(i,:));
    Model_mat{i} = cell2mat(Model(i,:));
end

% This iis the syntax for the percentile
P = prctile(Model_mat(1,:),(99))

Any help would be appreciated.

[Rik]

What exactly is your question? You seem to have a cell array of some sort, but you want the percentile function to do that conversion for you?

It would all be easier to understand if you attach your data to the question.

Answer 1

[Star Strider]

I am not certain what ‘Model’ is, however if it resembles this array, indexing into it as a cell array is straightforward —

Model = mat2cell(randn(10), ones(1,10),10)

Model = 10×1 cell array
    {[    0.3559 0.8681 0.5171 -1.9330 1.6688 -1.1222 -0.5230 0.4412 -2.6904 -0.3102]}
    {[ -0.4756 -0.0911 -0.4101 -1.2918 -0.8179 -0.6191 -0.2680 1.0212 0.5873 -2.4094]}
    {[    1.9852 0.2609 0.5998 -1.2972 -0.9622 1.9963 -0.4795 -0.4450 1.4728 -1.6797]}
    {[      1.7691 -0.3796 0.8273 0.1806 -0.7725 0.8785 0.3614 0.4027 -1.0187 0.6261]}
    {[       0.0359 1.2318 0.4570 1.4907 1.2762 0.3159 0.4088 -0.2682 0.9570 -0.1869]}
    {[-1.3947 -0.2354 2.6897 -0.0101 -1.2078 -0.1153 -1.3919 -1.0017 -1.7950 -2.0804]}
    {[  1.1646 -0.1035 -0.6771 0.6717 -0.3764 -0.5092 -1.9939 -0.9529 -0.5511 0.5268]}
    {[      -0.1635 1.9250 0.8539 0.1680 0.1291 0.5319 0.2604 -1.3130 -0.1796 1.1728]}
    {[     0.0794 0.6675 0.2235 1.7531 -0.9750 -0.3654 -0.1841 -0.5366 0.1033 0.6050]}
    {[    0.4417 -0.3038 0.6669 2.1762 -0.8061 -0.5986 2.7666 0.6785 -0.6522 -1.3446]}

for k = 1:size(Model,1)
    P(k) = prctile(Model{k,:},99);                              % Loop Approach
end

P

P = 1×10

1.6688 1.0212 1.9963 1.7691 1.4907 2.6897 1.1646 1.9250 1.7531 2.7666

P = prctile(cell2mat(Model),99,2)                               % Matrix Approach

P = 10×1

1.6688 1.0212 1.9963 1.7691 1.4907 2.6897 1.1646 1.9250 1.7531 2.7666

The loop approach would likely be appropriate if all the rows are not the same lengths. Otherwise, the matrix approach is likely more efficient.

.

[Alexander Guillen]

Thank you so much. I frequently look for MATLAB Answers for various applications, and often see you helping users. I want to say that your help is very valuable to the community.

[Star Strider]

As always, my pleasure!

I very much appreciate your compliment!