rename the pandas Series

Question

I have some wire thing when renaming the pandas Series by the datetime.date

import pandas as pd
a = pd.Series([1, 2, 3, 4], name='t')

I got a is:

0    1
1    2
2    3
3    4
Name: t, dtype: int64

Then, I have:

ts = pd.Series([pd.Timestamp('2016-05-16'), 
                pd.Timestamp('2016-05-17'), 
                pd.Timestamp('2016-05-18'), 
                pd.Timestamp('2016-05-19')], name='time')

with ts as:

0   2016-05-16
1   2016-05-17
2   2016-05-18
3   2016-05-19
Name: time, dtype: datetime64[ns]

Now, if I do:

ts_date = ts.apply(lambda x: x.date())
dates = ts_date.unique()

I got dates as:

array([datetime.date(2016, 5, 16), datetime.date(2016, 5, 17),
       datetime.date(2016, 5, 18), datetime.date(2016, 5, 19)], dtype=object)

---

I have two approaches. The wired thing is, if I do the following renaming (approach 1):

for one_date in dates:
    a.rename(one_date)
    print one_date, a.name

I got:

2016-05-16 t
2016-05-17 t
2016-05-18 t
2016-05-19 t

But if I do it like this (approach 2):

for one_date in dates:
    a = pd.Series(a, name=one_date)
    print one_date, a.name

2016-05-16 2016-05-16
2016-05-17 2016-05-17
2016-05-18 2016-05-18
2016-05-19 2016-05-19

---

My question is: why the method rename does not work (in approach 1)?

Answer 1

Because rename does not change the object unless you set the inplace argument as True, as seen in the docs.

Notice that the copy argument can be used so you don't have to create a new series passing the old series as argument, like in your second example.