使用 GroupBy 获取 Pandas 的平均值 - 获取数据错误：没有要聚合的数字类型 -
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/25397057/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Getting Average of Pandas with GroupBy- Getting DataError: No numeric types to aggregate -
提问by Max Song
I know that there are numerous questions about this, like Getting daily averages with pandasand How get monthly mean in pandas using groupbybut I'm getting a weird error.
我知道有很多关于这个的问题，比如使用Pandas获取每日平均值和如何使用 groupby 在Pandas中获取每月平均值，但我遇到了一个奇怪的错误。
Simple data set, with one index column (type timestamp) and one value column. Would like to get the monthly mean of the data.
简单的数据集，有一个索引列（类型时间戳）和一个值列。想获得数据的月平均值。
In [76]: df.head()
Out[76]:
A
2008-01-02 1
2008-01-03 2
2008-01-04 3
2008-01-07 4
2008-01-08 5
However, when I groupby, I get just the groups of the index and not of the value
但是，当我分组时，我只得到索引的组而不是值的组
In [74]: df.head().groupby(lambda x: x.month).groups
Out[74]:
{1: [Timestamp('2008-01-02 00:00:00'),
Timestamp('2008-01-03 00:00:00'),
Timestamp('2008-01-04 00:00:00'),
Timestamp('2008-01-07 00:00:00'),
Timestamp('2008-01-08 00:00:00')]}
Attempts to take means() result in an error:
尝试使用means()会导致错误：
Have tried both df.head().resample("M", how='mean')and df.head().groupby(lambda x: x.month).mean()
都试过df.head().resample("M", how='mean')和df.head().groupby(lambda x: x.month).mean()
and gets the error: DataError: No numeric types to aggregate
并得到错误： DataError: No numeric types to aggregate
In [75]: df.resample("M", how='mean')
---------------------------------------------------------------------------
DataError Traceback (most recent call last)
<ipython-input-75-79dc1a060ba4> in <module>()
----> 1 df.resample("M", how='mean')
/usr/local/lib/python2.7/site-packages/pandas/core/generic.pyc in resample(self, rule, how, axis, fill_method, closed, label, convention, kind, loffset, limit, base)
2878 fill_method=fill_method, convention=convention,
2879 limit=limit, base=base)
-> 2880 return sampler.resample(self).__finalize__(self)
2881
2882 def first(self, offset):
/usr/local/lib/python2.7/site-packages/pandas/tseries/resample.pyc in resample(self, obj)
82
83 if isinstance(ax, DatetimeIndex):
---> 84 rs = self._resample_timestamps()
85 elif isinstance(ax, PeriodIndex):
86 offset = to_offset(self.freq)
/usr/local/lib/python2.7/site-packages/pandas/tseries/resample.pyc in _resample_timestamps(self)
286 # Irregular data, have to use groupby
287 grouped = obj.groupby(grouper, axis=self.axis)
--> 288 result = grouped.aggregate(self._agg_method)
289
290 if self.fill_method is not None:
/usr/local/lib/python2.7/site-packages/pandas/core/groupby.pyc in aggregate(self, arg, *args, **kwargs)
2436 def aggregate(self, arg, *args, **kwargs):
2437 if isinstance(arg, compat.string_types):
-> 2438 return getattr(self, arg)(*args, **kwargs)
2439
2440 result = OrderedDict()
/usr/local/lib/python2.7/site-packages/pandas/core/groupby.pyc in mean(self)
664 """
665 try:
--> 666 return self._cython_agg_general('mean')
667 except GroupByError:
668 raise
/usr/local/lib/python2.7/site-packages/pandas/core/groupby.pyc in _cython_agg_general(self, how, numeric_only)
2356
2357 def _cython_agg_general(self, how, numeric_only=True):
-> 2358 new_items, new_blocks = self._cython_agg_blocks(how, numeric_only=numeric_only)
2359 return self._wrap_agged_blocks(new_items, new_blocks)
2360
/usr/local/lib/python2.7/site-packages/pandas/core/groupby.pyc in _cython_agg_blocks(self, how, numeric_only)
2406
2407 if len(new_blocks) == 0:
-> 2408 raise DataError('No numeric types to aggregate')
2409
2410 return data.items, new_blocks
DataError: No numeric types to aggregate
回答by FooBar
Yeah, you should try coercing Ato numeric with something like df['A'] = df['A'].astype(int). Might be worth checking if there's anything in the initial data read-in that caused it to be object instead of numeric as well.
是的，您应该尝试A使用类似df['A'] = df['A'].astype(int). 可能值得检查初始数据读入中是否有任何内容导致它也是对象而不是数字。
