I'm using MDS in julia. 'Here is my code

D = pairwise(Euclidean(), mat_new)
M = fit(MetricMDS, D; distances=true, maxoutdim=2, metric=isotonic)

And I got error saying

failure to converge after 300 iterations. Last change (0.1382513330677284) was greater than tolerance (0.001)

But if I use the code below instead

D = pairwise(Euclidean(), mat_new)
M = fit(MetricMDS, D; distances=true, maxoutdim=2)
Yte = predict(M)

It works fine.

I went through the document saying that "isotonic: converts dissimilarity values to ordinal disparities to perform non-metric MDS"

Does the "metric = isotonic" matter here? Can I say the second block of code is performing non-metric MDS?

0

There are 0 best solutions below