Scorecard and/or curve table should QC logs

[kwinkunks]

There should be some QC on the logs:

• Logs are the same as other logs (I'm looking at a well now where PEF == RHOB)
• Logs have unreasonable values (eg GR < 0)
• Logs are highly (i.e. ~perfectly) correlated: one is probably derived from the other

[EvanBianco]

np.isclose() is worth a try to see if two logs are very close, but we'd likely have to do some testing on whether the tolerances work for us.

But curves might be close and or correlated but offset element by element. We might need to do some kind of convolution or Toeplitz matrix shifted versions of the log and then check closeness.

http://docs.scipy.org/doc/numpy/reference/generated/numpy.isclose.html

[kwinkunks]

np.isclose() has two tolerance parameters, and another to tell it how to handle NaNs.

• rtol : float. The relative tolerance parameter.
• atol : float. The absolute tolerance parameter.
• equal_nan : bool. Whether to compare NaN’s as equal. If True, NaN’s in a will be considered equal to NaN’s in b in the output array.