Interpreting SVM calls and probs

[markdunning]

Hi,

Can you clarify please how to interpret the subtype calls and probabilities. In the following example from my dataset, Sample1 is assigned the Classical subtype, but the Proneural probability is clearly higher! I was thinking each sample would be assigned the class with the highest probability. Is this not the case?

Does this occur when two classes have very similar probability that it arbitrarily picks one?

Sample svm.subtype.call Classical Mesenchymal Proneural
Sample1 Classical 0.378 0.196 0.426
Sample2 Classical 0.395 0.223 0.382
Sample3 Classical 0.487 0.15 0.363

Many thanks,