timedelta与熊猫的奇怪结果

Question

I have a dataframe that looks like this:

df = pd.DataFrame({'date_sent': ['06/11/2015', '', 'Not required', '06/11/2015'], 
                   'date_published': ['06/11/2015', '', '', '23/01/2016']})

I want to calculate the difference between the two dates in each row, so first I convert the strings to date objects:

df.date_published =  pd.to_datetime(df.date_published.str.replace('Not required', ''))
df.date_sent =  pd.to_datetime(df.date_sent.str.replace('Not required', ''))

Then I subtract one from the other:

df['delay'] = df.date_published - df.date_sent

But this gives me peculiar results - it's not 226 days between 06/11/2015 and 23/01/2016:

df
  date_published  date_sent    delay
0     2015-06-11 2015-06-11   0 days
1            NaT        NaT      NaT
2            NaT        NaT      NaT
3     2016-01-23 2015-06-11 226 days

What am I doing wrong? I'm using pandas v0.18.

Answer 1

这些日期之间恰好是226天。

Answer 2

See if this helps.

   print pd.to_datetime('06/11/2016', dayfirst =True, format='%d/%m/%Y', errors = 'ignore')
   print pd.to_datetime('06/11/2016', format='%m/%d/%Y', errors = 'ignore')

2016-11-06 00:00:00
2016-06-11 00:00:00