anshanno anshanno - 3 months ago 28
Python Question

Better way to change pandas date format to remove leading zeros?

DataFrame look like:

OPENED
0 2004-07-28
1 2010-03-02
2 2005-10-26
3 2006-06-30
4 2012-09-21


I converted them to my desired format successfully but it seems very inefficient.

OPENED
0 40728
1 100302
2 51026
3 60630
4 120921


The code that I used for the date conversion is:

df['OPENED'] = pd.to_datetime(df.OPENED, format='%Y-%m-%d')
df['OPENED'] = df['OPENED'].apply(lambda x: x.strftime('%y%m%d'))
df['OPENED'] = df['OPENED'].apply(lambda i: str(i))
df['OPENED'] = df['OPENED'].apply(lambda s: s.lstrip("0"))

Answer

You can use str.replace, then remove first 2 chars by str[2:] and last remove leading 0 by str.lstrip:

print (type(df.ix[0,'OPENED']))
<class 'str'>
print (df.OPENED.dtype)
object

print (df.OPENED.str.replace('-','').str[2:].str.lstrip('0'))
0     40728
1    100302
2     51026
3     60630
4    120921
Name: OPENED, dtype: object

If dtype is already datetime use strftime and str.lstrip:

print (type(df.ix[0,'OPENED']))
<class 'pandas.tslib.Timestamp'>
print (df.OPENED.dtype)
datetime64[ns]

print (df.OPENED.dt.strftime('%y%m%d').str.lstrip('0'))
0     40728
1    100302
2     51026
3     60630
4    120921
Name: OPENED, dtype: object

Thank you Jon Clements for comment:

print (df['OPENED'].apply(lambda L: '{0}{1:%m%d}'.format(L.year % 100, L)))
0     40728
1    100302
2     51026
3     60630
4    120921
Name: OPENED, dtype: object