Cycle through data to find peaks/troughs and then section out the data points between them

Question

I have a script in Matlab which finds the peaks and valleys of my data set (looks kinda like a squished sine wave), eventually I want to average all of the data between each peak and trough. Here is the script I am working on, but I keep getting errors when I try to even select out the data between each peak and trough to build a matrix of just the data, even before getting to any of the averaging or anything like that. How can i get this script to work and what am I doing wrong? I get errors like:

The following error occurred converting from cell to double:
Error using double
Conversion to double from cell is not possible.

Error in Test (line 72)
             peakAverages(:,j) = {Peakdata((MaxIdx(j):MinIdx(j+1)),:)};

and if I change it to not be a cell array I get the opposite error. The code I am working on is:

for i = 1 : numel (tasknames) %for each test point do the math and find the peaks. tasknames lists the test points. 

  TaskData = cell2mat (task_data(i));

  Peakdata = TaskData(:, 4); %data is in the 4th column of my larger data matrix. 
  [Maxima, MaxIdx] = findpeaks (Peakdata, ...
                                'MinPeakHeight', mean (Peakdata), ...
                                'MinPeakDistance', 10);
  Maxima = Peakdata (MaxIdx);        

  Troughdata = 1.01 * max (Peakdata) - Peakdata;
  [Minima, MinIdx] = findpeaks (Troughdata, ...
                                'MinPeakHeight', mean (Troughdata), ...
                                'MinPeakDistance', 10);
  Minima = Peakdata (MinIdx); 

  MinLength = length (MinIdx); 
  MaxLength = length (MaxIdx); 

  %if there is a trough first, I want to take the first peak value
  %index and the second trough value index. Also since there are
  %different numbers of peaks and troughs, I want to make sure that
  %they match in length by always ending on a trough index. 

  %if there is a peak first, or MinIdx is greater than MaxIdx, do
  %normal j = 1 until whichver list of peaks/trouhgs is shorter. 
  if MinIdx(1) < MaxIdx(1)
    if MinLength > MaxLength
      for j = 1 : length (MaxIdx)
        peakAverages(:, j) = {Peakdata((MaxIdx(j) : MinIdx(j + 1)), :)};
      end
    elseif MinLength == MaxLength
      for j = 1 : length (MinIdx)
        peakAverages(:, j) = {Peakdata(MaxIdx(j) : MinIdx(j + 1), :)};
      end
    elseif MinLength < MaxLength
      for j = 1 : length (MinIdx)
        peakAverages(:, j) = {Peakdata(MaxIdx(j) : MinIdx(j + 1), :)};
      end
    end
  else 
    if MinLength > MaxLength
      for j = 1 : length (MaxIdx)
        peakAverages(j, :) = {Peakdata(MaxIdx(j) : MinIdx(j), :)};
      end
    elseif MinLength == MaxLength
      for j = 1 : length (MinIdx)
        peakAverages(:, j) = {Peakdata(MaxIdx(j) : MinIdx(j), :)};
      end
    elseif MinLength < MaxLength
      for j = 1 : length (MinIdx)
        peakAverages(:, j) = {Peakdata(MaxIdx(j) : MinIdx(j), :)};
      end
    end
end

This is the most horribly indented code I've seen in my entire life. Why?? — Tasos Papastylianou, Aug 13 '16 at 00:08
general advice: 1) When asking a question, ask it via a [Minimal Working Example](http://stackoverflow.com/help/mcve). 2) When writing code, avoid long spaghetti code like this, and think in a top-down way instead, such that the end result reads more like english via properly named functions. This not only helps readability, but isolates bugs more easily. — Tasos Papastylianou, Aug 13 '16 at 00:33
as for your error. what is `peakAverages` meant to be? An array of doubles? An array of cells? What's the "opposite error"? What do you expect `{Peakdata(MaxIdx(j) : MinIdx(j + 1), :)}` to return? (i.e. how many elements, and what type) — Tasos Papastylianou, Aug 13 '16 at 00:36
(just thought I'd clarify what I meant by [top-down design](https://en.wikibooks.org/wiki/A-level_Computing_2009/AQA/Problem_Solving,_Programming,_Data_Representation_and_Practical_Exercise/Problem_Solving/Top-down_design_and_Step-wise_refinement). ) — Tasos Papastylianou, Aug 13 '16 at 00:41
@TasosPapastylianou I honestly am not sure what it _needs_ to be but what I am trying to get is an array where each cell is columns of data that represent the data points on the downslope between each peak-trough found for each taskname. There always a peak followed by a trough in the data (unless the data collection started halfway between , hence all the if statements) so if there are 20peaks i would expect 20 columns of data for that task name with ~10 data points in each column. basically I want to break up the points that make up the downslope so I can perform functions on them. — user6711209, Aug 13 '16 at 00:45
eventually I would like to find the average of each of those downslope columns of data, thats why I called it peakaverages for now. — user6711209, Aug 13 '16 at 00:49

score 0 · Answer 1 · answered Aug 13 '16 at 01:35

0

Since you're not keen on creating a minimal working example reproducing the error to put things into context, I made a toy one myself:

% got rid of the for loop - not needed for a minimal example
Peakdata = sin([-50:50]).';
% remaining code is the same.

and I cannot reproduce your error. It just works.
Therefore your bug is in the definition of Peakdata or somewhere before that.

If you feel this is not true, then please provide a more appropriate example. I.e. something that is short (e.g. <10 lines of code) and reproducible, such that I could copy / paste onto my matlab session and run and check the error for myself. Otherwise the answer to your question is "no, it works, the bug is elsewhere".

answered Aug 13 '16 at 01:35

Tasos Papastylianou

21,371
2
28
57

I am not sure how I can without uploading a data file which I cannot do I dont think...I will try and get back shortly. – user6711209 Aug 13 '16 at 01:55
Peakdata is a 1210x1 double, I want to use the idx values to section it off, but I keep getting the above error when I try that. – user6711209 Aug 13 '16 at 01:59
Also if I try it with your sample data I get the error: Error using double Conversion to double from cell is not possible. – user6711209 Aug 13 '16 at 02:03
in that case the code you posted is not the same as the code you're running – Tasos Papastylianou Aug 13 '16 at 02:06
Tried it again and it works but I get the wrong output. why does peakAverages come out as a 1x8 cell array. it should only have one entry and in that array is 16 columns of data, since that is how many downslopes there are in that sine plot. – user6711209 Aug 13 '16 at 13:56

Cycle through data to find peaks/troughs and then section out the data points between them

1 Answers1