Hi all,
I have used the Decoding Toolbox (TDT) for a searchlight MVPA with a 4-fold leave-one-run-out-classification and was wondering about the scaling option. I have set the following parameters:
cfg.decoding.method = 'classification_kernel';
cfg.results.output = 'accuracy_minus_chance';
Which I am now realizing has invoked the default settings for scaling:
cfg.scale.method = 'min0max1global';
cfg.scale.estimation = 'all';
Is this the correct approach? What would be different if I set the scaling to ‘min0max1’?
Thanks a lot!
Jasmin