Description
From mailing list:
This is something that surprised me.
When reading from csv the values "True" and "False" are converted to bool, while "TRUE" and "FALSE"' to object
Using a converter {...: np.bool} I could take it into account.
Is it useful to take more possibilities into account knowing that there are a lot of possibilies True/False, Yes/No, ....?
In [19]: data = '''A;B
....: True;False
....: False;True'''
In [20]: df = pd.read_csv(StringIO(data), sep=';')
In [22]: df.dtypes
Out[22]:
A bool
B bool
In [23]: data = '''A;B
....: TRUE;FALSE
....: FALSE;TRUE'''
In [24]: df = pd.read_csv(StringIO(data), sep=';')
In [25]: df.dtypes
Out[25]:
A object
B object
In [27]: df = pd.read_csv(StringIO(data), sep=';', converters={'A':np.bool})
In [28]: df.dtypes
Out[28]:
A bool
B object