Emily Jenne, thank you for the feedback.

Yes, I believe SGT will be one good choice to apply on your data. From your example, it appears that the elements of the sequences are in the set {0, 1, …, 12}. This is a small set and you may not require to do dimension reduction with something like PCA. If the elements set is large, or your SGT feature space is sparse, you may still go for dimension reduction. Otherwise, you can use SGT features directly to build a classifier.