How to set a moving window

Hi everyone. I need help with a moving window creation: I'm analyzing behavioral data collected from rats and I need to create a moving window showing how their psycometric curves change through each session. I don't need help with the psycometric curves creation, but I really don't know how to set the loop to create a moving window with a window of 50 and a step of 1...

I need to do this with a LOOP, so I don't want to use any function (like movmean or conv for example)

I really hope you can help me because this is driving me crazy

Answer 1

[Image Analyst]

Why don't you want to use the built in function? Is it homework?

What do you want to do for elements 1-49? Or the final 49 elements? Do you want the window to shrink as your window would leave the array? Do you just want those places where the window fits complete in your array? Did you just try a simple for loop:

ratData = rand(1, 500); % Sample data
windowWidth = 50;
averagedData = zeros(1, length(ratData) - windowWidth);
for k = 1 : length(ratData) - windowWidth
	averagedData(k) = mean(ratData(k:k+windowWidth-1));
end
plot(ratData, 'b-', 'LineWidth', 2);
hold on;
plot(averagedData, 'r-', 'LineWidth', 2);
grid on;
legend('Actual', 'Averaged');
msgbox('Done!');

[Lisa Fontana]

Homework? Nono, I'm actually collecting data from rats (I'm in the lab right now).

I cannot use built function because my PI wants me to use a simple for loop because I've never done a moving window before.

For example.. I've done something similar with just 4 psycometric curves with this code:

bounds = 0:50:200;

mean_perf = nan (1, 4);

mean_Dec1 = nan (4, 9);

figure;

hold on;

cols2 = [0 0 1; 0 0.3 0.8; 0 0.4 0.7; 0 0.5 0.6];

pars2 = nan (4,4);

for iblock = 1: length(bounds)-1

block_index = data.trialID > bounds(iblock) & data.trialID <= bounds(iblock + 1) & data.waited;

mean_perf(1, iblock) = mean(data.hitmiss(block_index));

end

for iblock = 1: length (bounds)-1

block_index = data.trialID > bounds(iblock) & data.trialID <= bounds(iblock + 1) & data.waited;

for j = 1:length (stims)

meanDec1(iblock, j) = mean(data.action(stimID==stims(j) & ih1 & block_index));

count_Dec(iblock, j)= sum(stimID==stims(j) & ih1 & block_index);

end

end

for iblock = 1: length (bounds)-1

plot (stims,meanDec1(iblock,:), 'o','color',cols2(iblock,:),'markersize',4,'markerfacecolor',cols2(iblock,:));

pars2(iblock,:) = lsqcurvefit (@cumulativegaussianLapse, p0, stims, meanDec1(iblock,:));

prediction = cumulativegaussianLapse (pars2(iblock,:), xaxis);

plot (xaxis, prediction,'color',cols2(iblock,:),'linewidth',2);

end

title ('Performance trials 1-50, 51-100, 101-150, 151-200 per ih1');

xlabel ('Stims');

ylabel ('meanDec');

legend ('meanDec trials 1-50', 'Psycometric curve 1-50','meanDec trials 51-100', 'Psycometric curve 51-100','meanDec trials 101-150', 'Psycometric curve 101-150','meanDec trials 151-200', 'Psycometric curve 151-200', 'Location', 'southeast');

ylim ([0 1]);

I need to do the same thing but with many more psycometric curves, precisely with a moving window of 50 with a step of 1. I've tried many ways but I don't know why I can't figure it out, I have written much more complex code.. My idea is to set what happens in the first round of the loop, and then specify what happens from the second round on; I have the idea but I really don't know how to set the loop

[Image Analyst]

Maybe he doesn't know about the movmean() function so that's the reason why he suggested a loop. I see no real NEED for it to be a loop. Suggest the built-in movmean function to him. Or just go ahead and use it. I doubt he'd really care HOW you got it done.

Anyway, didn't my code do a manual 50 element moving mean? Not sure why it didn't meet your needs.

[Lisa Fontana]

No sorry I haven't tried your code yet, I'll do it right now.

I just wanted to explain you exactly what I ment. So do you think your code will work for what I want?

Answer 2

[Steven Lord]

Use the value of the for loop variable as the index of the first element in the window. Add a quantity to get the index of the last element in the window. Use linear indexing to extract the elements in the window.

x = (1:10).^2

x = 1×10

1 4 9 16 25 36 49 64 81 100

y = x(4:7) % [16 25 36 49]

y = 1×4

16 25 36 49