Вы можете использовать параметр date_parser
в read_csv
:
import sys
if sys.version_info.major<3:
from StringIO import StringIO
else:
from io import StringIO
import pandas as pd
temp=u"""date
201605
201606
201607
201608"""
dateparser = lambda x: pd.datetime.strptime(x, '%Y%m')
#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(StringIO(temp), parse_dates=[0], date_parser=dateparser)
print (df)
date
0 2016-05-01
1 2016-06-01
2 2016-07-01
3 2016-08-01
print (df.dtypes)
date datetime64[ns]
dtype: object