4 views (last 30 days)

Show older comments

Classical = pdist2(matrix_logdiff_nsample, mean(matrix_logdiff_nsample),'mahal');

p = size(matrix_logdiff_nsample,2);

chi2quantile = chi2inv(0.99,p);

[SFmcd, MFmcd, Fmcd, OutFmcd] = robustcov(matrix_logdiff_nsample);

plot(Classical, Fmcd, 'o')

line([chi2quantile, chi2quantile], [0, 120], 'color', 'r')

line([0, 80], [chi2quantile, chi2quantile], 'color', 'r')

hold on

plot(Classical(OutFmcd), Fmcd(OutFmcd), 'b*')

xlabel('Mahalanobis Distance')

ylabel('Robust Distance')

title('Distance Plot, Fast MCD method')

Adam Danz
on 12 Jul 2020

Edited: Adam Danz
on 13 Jul 2020

You just need to modify either of the two answers received for similar questions in the past (listed under your question).

To modify this answer, you just need to replace the centerPoint_x, centerPoint_y values with the coordinates you're using to define the quadrants.

Based on your image it would look something like,

centerPoint_x = 15; % approximate guess based on image

centerPoint_y = 18; % approximate guess based on image

Adam Danz
on 13 Jul 2020

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

Start Hunting!