TSData from Pandas

I have a time series classification problem (three classes A, B, and C). All of the series are in a single dataframe. I’m trying to follow the NATOPS ts classification example but I can’t get the data in the right format.

My data looks like this:
id, class, seq_id, var1, var2, … var56
001, A, 1, .34, .67,…, .098
001, A, 2, .30, .657,…, .0988

112, C, 210, .340, .765, …, .234

The the shape of the dataframe is thousands, by 59.
Any help in how I can reshape, modify the dataframe to create items is greatly appreciated.

Another item that I’m not sure if it will be an issue or not is that the classes have different lengths but there is also variation within class. Does that need to be standardized or padded to the longest lenght?
As and example class A has a mean length of 17 but varies from 12, 20 and class C has a mean length of 170 but varies from 150 to 220.