I am considering using LibSVM and I am creating the required feature vectors.
In almost all the example data there is an order for the features for example:
+1 1:3 2:1 3:5 4:2 10:8
Was wondering if this ordering is important, for example:
+1 4:2 10:8 2:1 1:3 3:5
Thanks
Although both of your notations are equivalent, LibSVM requires the Feature-IDs to be sorted ascendingly – I suppose this has I/O-reasons.