Python Dataframe 从几百万行的大日期时间索引中提取唯一日期列表

Question

My data frame has around 17 million rows. The index is DateTime. It is around one-second resolution one-year data. Now I want to extract a list of unique dates from it.

My code:

# sample df

df.index = DatetimeIndex(['2019-10-01 05:00:00', '2019-10-01 05:00:01',
               '2019-10-01 05:00:05', '2019-10-01 05:00:06',
               '2019-10-01 05:00:08', '2019-10-01 05:00:09',
               '2019-10-01 05:00:12', '2019-10-01 05:00:13',
               '2019-10-01 05:00:15', '2019-10-01 05:00:17',
               ...
               '2020-11-14 19:59:21', '2020-11-14 19:59:23',
               '2020-11-14 19:59:31', '2020-11-14 19:59:32',
               '2020-11-14 19:59:37', '2020-11-14 19:59:38',
               '2020-11-14 19:59:45', '2020-11-14 19:59:46',
               '2020-11-14 19:59:55', '2020-11-14 19:59:56'],
              dtype='datetime64[ns]', name='timestamp', length=17796121, freq=None)
dates = df.index.strftime('&Y-&m-%d').unique()

My above code gave the output. But it took around five minutes. Is there any better way by which I can get the dates much faster?

Answer 1

Save stftime for when you actually need the strings. It's pretty slow.

Try this:

dates = np.unique(dates.date)