No graph produced when contrast [1 1]

[ramBrain]

Dear Emin,

If I try and run the toolbox by setting the contrast as [1 1] (which should be an F-test, or ANOVA, right?) I get some accuracy score but no meaningful graph, just a matrix where all connections are weighted 1. Is this an expected behavior?

Thank you, Ramtin

[eminSerin]

Hi Ramtin,

Thank you for your interest in our toolbox. To use a network-based F-test as a feature selection method for regression problems, you will need to set the contrast as [0 1]. Please try running your analysis with this contrast and let me know how it goes.

Good luck with your analyses! Emin

[ramBrain]

Hi Emin,

Thank you for your answer.

I am trying to solve a classification problem, with two groups. I would need to run a two-tailed t-test (or an ANOVA). With the classic NBS the contrast would be [1 1] (F-test), is it not the same here?

Many thanks, Ramtin

[eminSerin]

Hi Ramtin,

I noticed there's a slight difference here. When dealing with classification problems, F-tests are not applicable, and instead, t-tests should be used. To use t-tests to select features, you'll need to set contrasts to [-1, 1] and [1, -1]. By doing this, you can determine the distinct contribution of edges that are more prominent in a particular group to the overall prediction performance.

Cheers, Emin