Kyle Beauchamp

Just the timestamp and methods associated with time shifts etc. We should just play with it and see if it's useful to us. I'm still not 100% sold.

We could create an object like that and use DataFrame.pivot_table() to reindex things as needed.

The idea of pivot_table is to take some "data" columns and convert them into the "index" for the object. I've used it in the context of dealing with NMR data....

I def think this is worth trying.

group_by is the key here.

Here's a half-working prototype. I still need to pivot the ass_i to use the timestamps as `index`, but it's otherwise working: ``` data = [] num_traj, traj_length = a.shape for...

OK, so one problem with my current approach is that `integer` series are incompatible with NANs (same as Numpy). Thus, we can't just throw NANs in.

I don't think this should be a problem, though.

Here's the updated version to index the individual trajectory assignments by their timestamps: ``` X = pd.DataFrame(data, columns=["traj","time","state"]) X["state"] for (k, ass_i) in X.groupby("traj"): ass_i = ass_i.pivot_table(rows=["time"])["state"] print(ass_i) ```

So by "hiding" the additional tutorials, perhaps what I mean is that we have a single link "Additional Tutorials" that brings users to a page with a list of tICA,...