ENH: GH11128 add weekday_name to DatetimeIndex and .dt #11813

bahrunnur · 2015-12-10T16:47:12Z

some examples:

In [5]: s2 = s.to_frame('value').assign(weekday=s.index.weekday_name)

In [6]: s2
Out[6]:
            value    weekday
2015-09-01      0    tuesday
2015-09-02      1  wednesday
2015-09-03      2   thursday
2015-09-04      3     friday
2015-09-05      4   saturday
2015-09-06      5     sunday
2015-09-07      6     monday
2015-09-08      7    tuesday
2015-09-09      8  wednesday
2015-09-10      9   thursday

and

In [7]: s3 = pd.Series(pd.date_range('2015-09-01', '2015-09-10'))

In [8]: s3.dt.weekday_name
Out[8]:
0      tuesday
1    wednesday
2     thursday
3       friday
4     saturday
5       sunday
6       monday
7      tuesday
8    wednesday
9     thursday
dtype: object

Please review my code 😄

jreback · 2015-12-10T17:01:57Z

pandas/tseries/tests/test_base.py

@@ -29,7 +29,7 @@ def setUp(self):
    def test_ops_properties(self):
        self.check_ops_properties(['year','month','day','hour','minute','second','weekofyear','week','dayofweek','dayofyear','quarter'])
        self.check_ops_properties(['date','time','microsecond','nanosecond', 'is_month_start', 'is_month_end', 'is_quarter_start',
-                                   'is_quarter_end', 'is_year_start', 'is_year_end'], lambda x: isinstance(x,DatetimeIndex))
+                                   'is_quarter_end', 'is_year_start', 'is_year_end', 'weekday_name'], lambda x: isinstance(x,DatetimeIndex))


can you add an explict test for the results of this (you can do it in tseries/test_timeseries.py,
e.g. create a datetimeindex and assert that i.weekday_name matches your expectation, e.g. since you are lowercasing

furthermore test with some NaT in there (these should be np.nan, which the masker will fix for you)

ok, I'm on it 💪

kawochen · 2015-12-10T17:05:09Z

The first letters should be capitalized?

jreback · 2015-12-10T17:08:32Z

pandas/tseries/index.py

+        """
+        weekdays = [calendar.day_name[i].lower() for i in range(7)]
+        return self._maybe_mask_results(_algos.arrmap_object(self.asobject.values, lambda x: weekdays[x.weekday()]))
+


You are going to want to do something like this as what you are doing is going to be much much slower (only reason is this is done for .date is that its creating python objects so no way around it).

In [25]: index = pd.date_range('20130101 09:00:00',periods=20) In [26]: weekdays = [calendar.day_name[i].lower() for i in range(7)] In [27]: pandas.core.common.take_nd(np.array(weekdays),index.weekday) Out[27]: array(['tuesday', 'wednesday', 'thursday', 'friday', 'saturday', 'sunday', 'monday', 'tuesday', 'wednesday', 'thursday', 'friday', 'saturday', 'sunday', 'monday', 'tuesday', 'wednesday', 'thursday', 'friday', 'saturday', 'sunday'], dtype='|S9')

then return this an an object array
you may need to do com._ensure_int64 on the i.weekday (I don't remember)

I'm totally noob about tslib.pyx because it is a Cython code that interacting with numpy, CPython datetime. I wanna give my best shot to update the implementation on the tslib.pyx. And then, acces it via _field_accessor in tseries/index.py with additional if branching checking if it is requesting a weekday_name. And we also can add like month_name perhaps? What do you think @jreback ?

jreback · 2015-12-10T17:09:20Z

pls also add this to Timestamp.

jorisvandenbossche · 2015-12-10T22:46:05Z

Sorry I did not comment on the original issue before you put effort in this, but I wonder a bit if this is worth a new method (we already have many many methods, so adding a new needs a strong case I think).
Wouldn't be some good docs be sufficient? Eg specifically add in the docstring that if you want to get the names instead of the numbers, how you can do that most easily?

jreback · 2015-12-11T00:39:24Z

I actually thing this is pretty reasonable. Its an adjust method for weekday that is useful. No harm IMHO at trivially expanding the API.

bahrunnur · 2015-12-11T02:25:03Z

@jorisvandenbossche that's ok, this is my first time doing PR on pandas and I learn a lot about the code and contributing process, etc. What I think about it is a 'nice to have' feature and it has Difficulty Novice label that I think this is a perfect issue for me to learn more about the underlying structure of pandas. IMHO I have not found a strong usage on this feature. But, @jreback may have another perspective about it.

What I think about the usage of it is on the REPL like IPython or just python to check thing that what we are intended to do on a date are doing exactly what we want. For example, in #11123 if we can see the day name seamlessly without interpreting the number into day name in our mind. Perhaps it could help our mind a little bit 😂

@kawochen Hmm yeah, I'm just thinking that the capitalized form may be useful for serialization into csv, excel, etc. Should I change to capitalized form @jreback? I need some advice to choose between lower case or capitalized 🙏.

jreback · 2015-12-11T02:34:05Z

prob should use what calendar gives back for the names

and, this actually could/should be implemented similarly to the _field_accessors (e.g. directly in cython).

jreback · 2016-01-06T17:20:07Z

@bahrunnur can you rebase / update

jreback · 2016-01-20T14:17:41Z

I still like this. @bahrunnur can you rebase / update

bahrunnur · 2016-01-22T10:51:08Z

Oh hi @jreback !
I'm sorry for the slow response. I've been on vacation since the first day of this year. Now, I have rebased my code.

jreback · 2016-01-22T20:26:55Z

pandas/tslib.pyx

+
+    if field == 'weekday_name':
+        for i in range(count):
+            if dtindex[i] == NPY_NAT: out[i] = -1; continue


put np.nan in if its NaT

needs some tests for NaT in the data as well (and test pd.NaT) as well

We're returning strings, isn't an empty string better for NaT? or str(np.nan)?
Where do we best put tests with NaT? Maybe just add a case to an existing test? How do I run these tests, where is the beginning of all this spagetti? (bear with me, I'm on a steep curve).

no np.nan is the empty indicator in non-datetimelike objects.

tests go in tseries/tests_timeseries.py look where things like is_year_end properties are tested

jreback · 2016-01-30T15:21:55Z

can you update

jreback · 2016-03-12T17:51:04Z

can you rebase/update

jreback · 2016-04-05T19:13:15Z

closed in favor of #12803

closes #11128 replaces #11813 - [x] tests added / passed - [x] passes ``git diff upstream/master | flake8 --diff`` - [x] whatsnew entry Completed merge of Bahrunnur's PR 'weekday-name' Moved 'enhancements' to v0.18.1.txt Added tests to pandas/tests/series/test_datetime_values.py Added weekday_name property producing 'NaT' to NaTType class Author: Bastiaan <[email protected]> Author: Bahrunnur <[email protected]> Closes #12803 from BastiaanBergman/ENH11128 and squashes the following commits: c579d71 [Bastiaan] Tiny fixes as requested by jreback, GH12803. 6f246d5 [Bastiaan] Small fixes as requested by jreback, GH12803. 7b14d5c [Bahrunnur] ENH: GH11128 add weekday_name to DatetimeIndex and .dt

jreback added this to the 0.18.0 milestone Dec 10, 2015

jreback added Enhancement Datetime Datetime data dtype labels Dec 10, 2015

jreback reviewed Dec 10, 2015
View reviewed changes

bahrunnur force-pushed the weekday-name branch 3 times, most recently from 9142992 to f5bb183 Compare December 11, 2015 16:34

bahrunnur force-pushed the weekday-name branch from f5bb183 to e652e91 Compare January 22, 2016 10:47

ENH: GH11128 add weekday_name to DatetimeIndex and .dt

e652e91

jreback reviewed Jan 22, 2016
View reviewed changes

jreback modified the milestones: Next Major Release, 0.18.0 Feb 2, 2016

jreback removed this from the Next Major Release milestone Mar 13, 2016

This was referenced Mar 30, 2016

ENH: add weekday_name to DatetimeIndex and .dt #11128

Closed

ENH: GH11128 add weekday_name to DatetimeIndex and .dt #12803

Closed

jreback closed this Apr 5, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ENH: GH11128 add weekday_name to DatetimeIndex and .dt #11813

ENH: GH11128 add weekday_name to DatetimeIndex and .dt #11813

bahrunnur commented Dec 10, 2015

jreback Dec 10, 2015

jreback Dec 10, 2015

bahrunnur Dec 11, 2015

kawochen commented Dec 10, 2015

jreback Dec 10, 2015

bahrunnur Dec 11, 2015

jreback commented Dec 10, 2015

jorisvandenbossche commented Dec 10, 2015

jreback commented Dec 11, 2015

bahrunnur commented Dec 11, 2015

jreback commented Dec 11, 2015

jreback commented Jan 6, 2016

jreback commented Jan 20, 2016

bahrunnur commented Jan 22, 2016

jreback Jan 22, 2016

jreback Jan 22, 2016

BastiaanBergman Apr 1, 2016

jreback Apr 1, 2016

jreback Apr 1, 2016

jreback commented Jan 30, 2016

jreback commented Mar 12, 2016

jreback commented Apr 5, 2016

ENH: GH11128 add weekday_name to DatetimeIndex and .dt #11813

ENH: GH11128 add weekday_name to DatetimeIndex and .dt #11813

Conversation

bahrunnur commented Dec 10, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kawochen commented Dec 10, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Dec 10, 2015

jorisvandenbossche commented Dec 10, 2015

jreback commented Dec 11, 2015

bahrunnur commented Dec 11, 2015

jreback commented Dec 11, 2015

jreback commented Jan 6, 2016

jreback commented Jan 20, 2016

bahrunnur commented Jan 22, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Jan 30, 2016

jreback commented Mar 12, 2016

jreback commented Apr 5, 2016