Separating of succecive monotonic vectors

Hi every one,

I have a vector which contains values. These values consist of several monotonic vectors (as show in the figure). For this purpuse, i'd like to extract each monotone vector alone. Is there a solution to be able to isolate each monotone vector from the global one?

Thank you for your attention and tyour help

[the cyclist]

Could you also upload the original data from which you made that plot, in a MAT file?

[moncef soualhi]

Hi, yes with pleasure. Please find in attachment the .mat file data.

Thank you very much for your help

Answer 1

[Fabio Freschi]

Let's assume your data are stored in x and y vectors. I take them from your picture

% get the data from picture
fig = gcf;
axObjs = fig.Children
dataObjs = axObjs.Children
x = dataObjs(1).XData;
y = dataObjs(1).YData;

your data are noisy. Let's specify a suitable threshold

threshold = 200;

the "negative jumps can be identified using diff

idx = [0 find(diff(y) < -threshold) length(y)];

the chunks of monotonical data have different sizes, I load them into a cell array

X = arrayfun(@(i)x(idx(i)+1:idx(i+1)),1:length(idx)-1,'UniformOutput',false);
Y = arrayfun(@(i)y(idx(i)+1:idx(i+1)),1:length(idx)-1,'UniformOutput',false);

Graphical check

figure, hold on
for i = 1:length(Y)
    plot(X{i},Y{i})
end

[moncef soualhi]

Hi,

Thank you very much, it's work well. However, i have a question if you can answer me, please.

In real, there is 25 monotonic vectors. with your script, it found that there is 26 monotonic vectors. Hiwever, when i change the threshold from 200 to 400 the results was exactly crorrect as the source data separated.

Could please explain me these line below, please:

idx = [0 find(diff(y) < -threshold) length(y)];

Y = arrayfun(@(i)y(idx(i)+1:idx(i+1)),1:length(idx)-1,'UniformOutput',false);

Thank you very much for your assistance and your help

[Fabio Freschi]

diff(y) < -threshold is a logical vector idx that contains the indices in the data vector where the jump is larger than the thrashold. With find we get the indices values. I added a leading 0 and a trailing 7538 (vector length) for the successive processing.

arrayfun is a nice matlab function. Have a look at the documentation for details. In this case that mimics in one row the following for loop

for i = 1:length(idx)-1
    Y{i} = y(idx(i)+1:idx(i+1));
end

It loads in the ith cell of Y the monotonic portion of the vector y with starting index idx(i)+1 and final index idx(i+1)

[moncef soualhi]

Thank you, it's very help full