Finding a significant difference between two datasets

16 views (last 30 days)
I would like to find spikes in a dataset that are significantly different from a second dataset.
The picture attached shows two datasets and I would like to find the spike at 255 on the x-axis.
It would be helpful if I find these significant spikes automatically with a few commands.
I think it could be possible that I have to apply a tolerance or something that says only find data that has a greater difference than such and such (depending on the datasets that I compare).
So far I only could find the overall maximum y-value and its location but that's not what I want to find.
The output should be something like:
y = [max difference between two data sets]
x = [value at which significant difference occurs]

Accepted Answer

John Chilleri
John Chilleri on 10 Mar 2017
Edited: John Chilleri on 10 Mar 2017
Given data vectors a and b of equal length, you can use the following (assuming your x values are the default 1,2,...):
x = 1:length(a);
y = abs(b-a);
[y, ind] = sort(y,'descend');
x = x(ind);
Now the vector y contains the differences sorted from largest to smallest, and the x vector contains the corresponding x values.
Hope this helps!

More Answers (1)

Steven Lord
Steven Lord on 10 Mar 2017
If you're using release R2017a take a look at the isoutlier function.


Find more on Shifting and Sorting Matrices in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!