# Range features with SVM

I have some data that I'd like to apply classification to:

Label    Start1, End1,   Start2, End2,    Start3, End3
0       range 1         range 2          range 3
1 ...
2 ...
1 ...
...


SVM seems like a good approach for this particular type of data, but I'm not very confident with selecting features. The natural features are the ranges above, expressed as a start and end value, and classification seeks to find the label for particular value that falls within a range (any range). Should there be 6 features, or would it perhaps be better to use Start and Length?

Thanks

-
Is there a reason why you think SVM is particularly suitable for this type of data? Are there some data samples you can post ? –  image_doctor May 9 '12 at 8:39
You have to compare apples to apples. Each instance in the training dataset needs to look just like the instance you are trying to classify. It's not clear to me that you are doing that. –  karenu May 9 '12 at 21:49