Using Pandas within Pymc3 context?

[chartl]

You can always try to unwrap into an array

xxdat_ad = np.array([float(u) for u in xxdat['ad']])

and see where the error occurs.