how to use time to index in a datetimeIndex array

Question

I have a pandas dataframe which has datetime as its index (shown below)

              ['2018-02-13 11:55:00', '2018-02-13 12:00:00',
               '2018-02-13 12:05:00', '2018-02-13 12:10:00',
               '2018-02-13 12:15:00', '2018-02-13 12:20:00',
               '2018-02-13 12:25:00', '2018-02-13 12:30:00',
               '2018-02-13 12:35:00', '2018-02-13 12:40:00',
               ...
               '2018-02-19 04:40:00', '2018-02-19 04:45:00',
               '2018-02-19 04:50:00', '2018-02-19 05:05:00',
               '2018-02-19 05:10:00', '2018-02-19 05:15:00',
               '2018-02-19 05:20:00', '2018-02-19 05:25:00',
               '2018-02-19 05:30:00', '2018-02-19 05:40:00'])

I would like to filter all the value which has time 12:00:00 for example, therefore I am looking for a return value of

['2018-02-13 12:00:00','2018-02-14 12:00:00','2018-02-15 12:00:00','2018-02-16 12:00:00','2018-02-17 12:00:00', '2018-02-18 12:00:00']

How can I perform such indexing please?

cs95 · Accepted Answer · 2018-10-01 22:35:54Z

6

Using the hour attribute is not enough if you want to find entries corresponding to noon at 12:00PM. Instead, compare your index/date-range with a datetime object:

dt = pd.date_range('2018-02-13', '2018-02-19', freq='1h')
dt[dt.time == datetime.time(12)]

DatetimeIndex(['2018-02-13 12:00:00', '2018-02-14 12:00:00',
               '2018-02-15 12:00:00', '2018-02-16 12:00:00',
               '2018-02-17 12:00:00', '2018-02-18 12:00:00'],
              dtype='datetime64[ns]', freq=None)

If you want to introduce a minutes or seconds component, simply change datetime.time(12) to datetime.time(12, 5) for 12:05:00PM, or datetime.time(12, 5, 30) for 12:05:30PM, and so on.

edited Oct 1, 2018 at 22:35

answered Oct 1, 2018 at 22:22

cs95

406k106 gold badges744 silver badges797 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Victor Over a year ago

Thanks. However, it does not work on my data. Using data.time == datetime(12,5) it reports no time attribute in dataframe data, using data.index == datetime(12,5) it returns an empty dataframe, using data.date_Time == datetime(12,5) it reports no time attribute as well. My index's name however is Date_Time

cs95 Over a year ago

@Victor Do df.index.time == datetime.time(12, 5)

Victor Over a year ago

Thanks again, so which one performs faster? or a more preferable in python please? (data[(data.index.minute==5) & (data.index.hour == 10)] or data[data.index.time == datetime.time(12,5)])

cs95 Over a year ago

@Victor The former requires two masks. I think the latter would be better, but the best is to test it and check. If you wanted to introduce a seconds component, you will need to introduce a 3rd mask (df.index.seconds == 30) and & it with the first two masks. For the second method, just add "30" to the datetime.time call.

Victor Over a year ago

I agree with you, if second is introduced the code will look more complicated. I recall that I have seen some tutorial of creating a timer for time measurement, let me do this research now. Many thanks for your quick help!

cs95 · Accepted Answer · 2018-10-01 22:35:19Z

2

You can filter using strftime

df[df.index.strftime('%H:%M:%S') == '12:00:00']

edited Oct 1, 2018 at 22:35

cs95

406k106 gold badges744 silver badges797 bronze badges

answered Oct 1, 2018 at 22:23

Vaishali

38.5k5 gold badges62 silver badges88 bronze badges

Collectives™ on Stack Overflow

how to use time to index in a datetimeIndex array

2 Answers 2

5 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

5 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related