Pandas read dataframe from csv with index as string, not int

Question

My csv file is as following :

INDEX, VAL
04016170,22
04206261,11
0420677,11

df = pd.read_csv('data.csv', index_col='INDEX')

How can I force pandas to read the index as string and not as integer (to preserve the first 0) ?

Answer 1

You can pass the dtype as a param this will map the column to the passed dtype:

In [130]:
import io
import pandas as pd
t="""INDEX,VAL
04016170,22
04206261,11
0420677,11"""
df = pd.read_csv(io.StringIO(t), index_col='VAL', dtype={'INDEX':str})
df

Out[130]:
        INDEX
VAL          
22   04016170
11   04206261
11    0420677

In [131]:    
df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 3 entries, 22 to 11
Data columns (total 1 columns):
INDEX    3 non-null object
dtypes: object(1)
memory usage: 48.0+ bytes

EDIT

OK, you can do it this way, there is a bug here when you explicitly set the index_col in read_csv, so you have to load the csv in first and then call set_index after loading:

In [134]:
df = pd.read_csv(io.StringIO(t), dtype={'INDEX':str})
df = df.set_index('INDEX')
df

Out[134]:
          VAL
INDEX        
04016170   22
04206261   11
0420677    11