根据索引分配序列值

Question

having a simple series:

>>> s = pd.Series(index=pd.date_range('2016-09-01','2016-09-05'))
>>> s

2016-09-01   NaN
2016-09-02   NaN
2016-09-03   NaN
2016-09-04   NaN
2016-09-05   NaN
Freq: D, dtype: float64

Am I able to set series values based on its index? Let's say, I want to set series values to dayofweek of corresponding index entry. Of course, I can accomplish it easily by constructing series from scratch:

>>> dr = pd.date_range('2016-09-01','2016-09-05')
>>> s = pd.Series(data=dr.dayofweek, index=dr)
>>> s

2016-09-01    3
2016-09-02    4
2016-09-03    5
2016-09-04    6
2016-09-05    0
Freq: D, dtype: int32

I am also able to accomplish it using dataframe: df['old_column'] = df.index.dayofweek . Is it possible to set series in similar manner (using the only "column" series have)? .apply() and .map() methods seem as no help, since they do not work with index values...

Answer 1

You can do it like this:

s[s.index] = s.index.dayofweek

s
Out: 
2016-09-01    3
2016-09-02    4
2016-09-03    5
2016-09-04    6
2016-09-05    0
Freq: D, dtype: int32

Answer 2

When using apply on a series, you cannot access the index values. However, you can when using apply on a dataframe. So, convert to a dataframe first.

s.to_frame().apply(lambda x: x.name.dayofweek, axis=1)

2016-09-01    3
2016-09-02    4
2016-09-03    5
2016-09-04    6
2016-09-05    0
Freq: D, dtype: int64

This is a demonstration of how to access the index value via apply . If assigning a column to be the dayofweek values is the only objective, s.index.dayofweek is far more appropriate.