Comments on: How to draw a covariance error ellipse? https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/ A blog about intelligent algorithms, machine learning, computer vision, datamining and more. Fri, 21 Jul 2017 05:50:18 +0000 hourly 1 https://wordpress.org/?v=3.8.39 By: Rick Sprague https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-397 Mon, 12 Jun 2017 02:02:31 +0000 http://www.visiondummy.com/?p=288#comment-397 My half minor axis is a negative value and the square root returns a “nan”.

]]> By: brian https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-381 Wed, 24 May 2017 17:00:25 +0000 http://www.visiondummy.com/?p=288#comment-381 I don’t think you need to monkey around with rotations and what not. I think you can just do
avg’ + chisquare_val * (sqrt(eigenval(1,1))*eigenvec(:,1)*cos(theta_grid) + sqrt(eigenval(2,2))*eigenvec(:,2)*sin(theta_grid));
once you have defined chisquare_val, avg, theta_grid, and eigenval and eigenvec.

]]> By: Dani C. https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-380 Thu, 18 May 2017 12:56:24 +0000 http://www.visiondummy.com/?p=288#comment-380 Hi! many thanks for this helpful post: just one more question about this code (that I’m using too).
According to what you are saying, the variable j correspond to the z-score: is it correct? then for 95% CI we should use j=1.96.
Why we shoulld use z-score instead of Chi (as you explained in your post)?

Many Thanks

]]> By: Guilherme https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-371 Tue, 11 Apr 2017 17:46:25 +0000 http://www.visiondummy.com/?p=288#comment-371 Hello!! Thanks a lot for the code!
How can I calculate the area of the drawned error ellipse?

]]> By: Julean https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-368 Wed, 05 Apr 2017 04:36:38 +0000 http://www.visiondummy.com/?p=288#comment-368 Hi Vincent, what a great article, thank you very much.
Regarding your comment, would you elaborate on that as I’m not getting what you meant. Based on my understanding, all point lie on the same ellipse have the same Mahalanobis distance and consequently Chi-square probability. Is that right?

]]> By: DS https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-353 Wed, 08 Mar 2017 13:46:14 +0000 http://www.visiondummy.com/?p=288#comment-353 Hi,
Great tutorial!
Aren’t the eigenvectors given in the columns?
So why is it in the MATLAB code that the smallest_eigenvec = eigenvec(1,:); instead of smallest_eigenvec = eigenvec(:,1)?

Thanks!

]]> By: Marta https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-334 Tue, 20 Dec 2016 19:44:37 +0000 http://www.visiondummy.com/?p=288#comment-334 Vincent – thanks a lot for your article. It become clear for me right now how it works for normal distribution. But what if your data are not normal? What would you do than?

]]> By: pana https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-329 Thu, 24 Nov 2016 15:11:10 +0000 http://www.visiondummy.com/?p=288#comment-329 Great Work! :)
What about the case that one of our datasets are non normal distributed?
Thank you in advance!

]]> By: pana https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-327 Thu, 17 Nov 2016 09:06:12 +0000 http://www.visiondummy.com/?p=288#comment-327 Hi! Great work also for you! :)

Do you know the case where one of the data y1, y2 or y3 are non normal?

Thank you in advance!

]]> By: Paola Zampino https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/#comment-326 Mon, 14 Nov 2016 16:51:52 +0000 http://www.visiondummy.com/?p=288#comment-326 Hi, I have to evaluate the eccentricity of the ellipse, so I need the value of each axis, how can i find them? maybe they are the same as eigenvalues?!

]]>