MNT/TST: remove xcorr and acorr from test_datetime #27427
Conversation
6a16664 to
99fab6a
Compare
If we claim to support it, i.e. this participates in units, then it should probably be tested. Also timedeltas are common for these https://online.stat.psu.edu/stat462/node/188/ |
|
Xcorr and acorr do not participate in units. Both work by assuming x and y are spaced equally in time (or space). You can of course assign units to the lags, but we do not try to do this in xcorr/acorr. |
It might be good to generate a table of what participates in units and what doesn't, possibly added either to https://matplotlib.org/devdocs/users/explain/axes/axes_units.html#more-about-unit-support or https://matplotlib.org/devdocs/api/units_api.html |
lib/matplotlib/axes/_axes.py
Outdated
| @@ -2055,7 +2056,9 @@ def xcorr(self, x, y, normed=True, detrend=mlab.detrend_none, | |||
|
|
|||
| Parameters | |||
| ---------- | |||
| x, y : array-like of length n | |||
| x, y : array-like of length n. Neither x nor y are run through | |||
There was a problem hiding this comment.
| x, y : array-like of length n. Neither x nor y are run through | |
| x, y : array-like of length n. Neither *x* nor *y* are run through |
07ed125 to
1f37e49
Compare
1f37e49 to
cb5463d
Compare
cb5463d to
403873f
Compare
2b26a0a to
7cfd688
Compare
Currently
xcorrandacorrare intest_datetimes.py. However, I think it is highly unlikely anyone would want to pass datetimes in asxvalues to either of these methods. It is also not clear what giving the unit conversion means for these arguments. Modified the docstrings to make this clear.