DEPR: fillna method kwd #53496
DEPR: fillna method kwd #53496
Conversation
| .. deprecated:: 2.1.0 | ||
| Use obj.ffill or obj.bfill instead. |
There was a problem hiding this comment.
Since we're going to deprecate the method argument, we should just deprecate all of fillna here; using value with groupby provides no benefit over using it in e.g. DataFrame.fillna. I was already leaning this way while I was working on #53438.
There was a problem hiding this comment.
Good point, will update.
There was a problem hiding this comment.
would you be OK with just doing the method deprecation here and a separate PR to deprecate the rest of the method?
pandas/core/resample.py
Outdated
| @@ -823,6 +823,14 @@ def fillna(self, method, limit: int | None = None): | |||
| 2018-01-01 01:30:00 6.0 5 | |||
| 2018-01-01 02:00:00 6.0 5 | |||
| """ | |||
| if method is not None: | |||
There was a problem hiding this comment.
Unlike other fillna methods, there is no value argument here, so we'd deprecate the entire method instead. But there is also a nearest option which you can't get with ffill/bfill. Does it make sense to add an nfill?
There was a problem hiding this comment.
does interpolate do that?
There was a problem hiding this comment.
That's a bit different; fillna does not fill existing NA values, only those created by upsampling.
ser = pd.Series(
data=[1, np.nan, 3],
index=[
dt.datetime(2023, 3, 1, 7, 0, 0),
dt.datetime(2023, 3, 1, 7, 0, 2),
dt.datetime(2023, 3, 1, 7, 0, 4),
]
)
print(ser.resample("1s").interpolate("nearest"))
# 2023-03-01 07:00:00 1.0
# 2023-03-01 07:00:01 1.0
# 2023-03-01 07:00:02 1.0
# 2023-03-01 07:00:03 3.0
# 2023-03-01 07:00:04 3.0
# Freq: S, dtype: float64
print(ser.resample("1s").fillna(method="nearest"))
# 2023-03-01 07:00:00 1.0
# 2023-03-01 07:00:01 NaN
# 2023-03-01 07:00:02 NaN
# 2023-03-01 07:00:03 3.0
# 2023-03-01 07:00:04 3.0
# Freq: S, dtype: float64
This does seem to me to be useful when upsampling on a time series with irregular intervals.
There was a problem hiding this comment.
you've looked at this more closely than i have. what's your suggestion here?
There was a problem hiding this comment.
Whoops, there is already a .nearest! We're good here; deprecating the entire method doesn't lose any functionality.
pandas/core/resample.py
Outdated
| warnings.warn( | ||
| f"{type(self).__name__}.fillna is deprecated and will be removed " | ||
| "in a future version. Use obj.ffill(), obj.bfill(), " | ||
| "or obj.neaerest() instead.", |
|
Thanks @jbrockmendel |
* DEPR: fillna method kwd * deprecate Resampler.fillna entirely * suppress doctest warnings * typo fixup * doctest * mypy fixup * lint fixup
This PR deprecates `method` parameter in all public `fillna` APIs to match pandas: pandas-dev/pandas#53496 This PR: ``` = 1056 failed, 99098 passed, 2069 skipped, 776 xfailed, 312 xpassed, 20 errors in 670.87s (0:11:10) = ``` On `pandas_2.0_feature_branch`: ``` = 1584 failed, 98570 passed, 2069 skipped, 776 xfailed, 312 xpassed, 20 errors in 737.24s (0:12:17) = ```
doc/source/whatsnew/vX.X.X.rstfile if fixing a bug or adding a new feature.