Calculating the mean per row of a nested cell arrays and/or RMSE of cell arrays

Question

ErikaZ 2018-6-30

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/408146-calculating-the-mean-per-row-of-a-nested-cell-arrays-and-or-rmse-of-cell-arrays

回答： ErikaZ 2018-7-5

I have a cell array:

A =   3×1 cell array
    [14×122 double]
    [14×395 double]
    [14×424 double]

And cell:

B =  3×1 cell array
    [14×122 double]
    [14×395 double]
    [14×424 double]

I need to calculate the RMSE of the first “within” row of A and the odd “within” rows of B. I am using the following equation:

RMSE = sqrt(mean((y - yhat).^2));

I have working:

cellfun(@(p) power(p,2),(cellfun(@(x,y) x(1,:) - y(1:2:end,:), test_Targets, Predicted_Targets,'uni', false)),'uni',false);

Which gives me the squared values of the difference of the rows I need. This returns a

ans =   3×1 cell array
    [7×122 double]
    [7×395 double]
    [7×424 double]

Then, I am stuck at trying to calculate the mean. Technically, I need to do mean(C{i}(j,:)) but I was not have to implement it. I would like to have just one line for the whole RMSE. I tried

C=cellfun(@(x) power(x,2),(cellfun(@(x,y) x(1,:) - y(1:2:end,:), test_Targets, Predicted_Targets,'uni', false)),'uni',false);
j =cellfun(@(x) transpose(1:size(x,1)/2), Predicted_Targets,'uni',false);
cellfun(@(v,w) mean(v(w,:)), C,j,'uni',false)
ans = 3×1 cell array
    [1×122 double]
    [1×395 double]
    [1×424 double]

Which is wrong because I am expecting a 3x1 containing 7x1 values.

At the end I would like to have something like:

RMSE = sqrt(arrayfun(@mean,cellfun(@(x) power(x,2),(cellfun(@(x,y) x(1,:) - y(1:2:end,:), test_Targets, Predicted_Targets,'uni', false)),'uni',false),'uni', false));

(This code is also wrong. It was my initial try before trying to implement the j. I also tried immse since it was suggested in another answer).

Thank you

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

dpb 2018-6-30

OK, so what is the actual computation you're trying to do; I got lost in the missteps along the way. I think if you illustrate with small subset the calculation and show the result for the sample data and how to get it (mathematically, not necessarily by code) folks will have better chance to write a solution.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

ErikaZ 2018-7-5

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/408146-calculating-the-mean-per-row-of-a-nested-cell-arrays-and-or-rmse-of-cell-arrays#answer_327684

在 MATLAB Online 中打开

I had to use 2 for loops, I couldn't figure it out with cellfun and/or arrayfun.

      % Individual RMSE for angle and moment
      C=cellfun(@(x) power(x,2),(cellfun(@(x,y) x(1,:) - y(1:2:end,:), test_Targets, Predicted_Targets,'uni', false)),'uni',false);
      B=cellfun(@(x) power(x,2),(cellfun(@(x,y) x(2,:) - y(2:2:end,:), test_Targets, Predicted_Targets,'uni', false)),'uni',false);
      for i=1:3
          for j=1:7
              RMSE_angle{i,1}(j,1)=sqrt(mean(C{i,1}(j,:)));
              RMSE_moment{i,1}(j,1)=sqrt(mean(B{i,1}(j,:)));
          end
      end