使用'datetime64 [ns]'格式从熊猫数据框中提取

Question

I have a dataframe which has elements as:

df1[1:4]
                   Sims
2014-01-02  [51, 53, 51, 3...
2014-01-03  [56, 48, 64, ...
2014-01-04  [57, 45, 47, ...

The sims are list of 500 elements each.

I have another dataframe as:

df2[1:4]
                          Date  Month  Day   HE    Year
DateTime                                               
2012-01-01 02:00:00 2012-01-01    1.0  1.0  2.0  2012.0
2012-01-01 03:00:00 2012-01-01    1.0  1.0  3.0  2012.0
2012-01-01 04:00:00 2012-01-01    1.0  1.0  4.0  2012.0

I am trying the following in various configurations:

df1[df2['Date']]

But it errors out complaining about time format difference between df1 index and df2['Date'] . However, both have same time format as shown below.

df1.index[1:4]
DatetimeIndex(['2014-01-02', '2014-01-03', '2014-01-04'], dtype='datetime64[ns]', freq=None)


    df2['Date'][1:4].values
array(['2012-01-01T00:00:00.000000000', '2012-01-01T00:00:00.000000000',
       '2012-01-01T00:00:00.000000000'], dtype='datetime64[ns]')

How do I make the following work:

df1[df2['Date']]

Edit: Error message:

KeyError: "['2012-01-01T00:00:00.000000000' '2012-01-01T00:00:00.000000000'\n '2012-01-01T00:00:00.000000000' ..., '2016-12-31T00:00:00.000000000'\n '2016-12-31T00:00:00.000000000' '2016-12-31T00:00:00.000000000'] not in index"

Answer 1

df1[df2['Date']] -type indexing tends to error in my experience if you are trying to index on rows instead of columns. The problem is presumably that you let pandas guess over which axis you whish to slice, and this doesn't always pan out as desired.

You could try using a more explicit indexing method such as df1.loc[df2['Date'], :] or df1.xs(df2['Date'], 0) .