BUG: ValueError: 1.0 and 0.0 cannot be cast to bool when using pd.read_csv()

[aauss]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Code Sample

import io
import pandas as pd

example_csv = """
boolean_column, string_column
1.0, a
0.0, b
,c"""

csv = io.StringIO(example_csv)
df = pd.read_csv(csv, dtype={"boolean_column": "boolean"})

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-2-c698a4bcbc2d> in <module>
      1 csv = io.StringIO(example_csv)
----> 2 df = pd.read_csv(csv, dtype={"boolean_column": "boolean"})

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\io\parsers.py in parser_f(filepath_or_buffer, sep, delimiter, header, names, index_col, usecols, squeeze, prefix, mangle_dupe_cols, dtype, engine, converters, true_values, false_values, skipinitialspace, skiprows, skipfooter, nrows, na_values, keep_default_na, na_filter, verbose, skip_blank_lines, parse_dates, infer_datetime_format, keep_date_col, date_parser, dayfirst, cache_dates, iterator, chunksize, compression, thousands, decimal, lineterminator, quotechar, quoting, doublequote, escapechar, comment, encoding, dialect, error_bad_lines, warn_bad_lines, delim_whitespace, low_memory, memory_map, float_precision)
    674         )
    675 
--> 676         return _read(filepath_or_buffer, kwds)
    677 
    678     parser_f.__name__ = name

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\io\parsers.py in _read(filepath_or_buffer, kwds)
    452 
    453     try:
--> 454         data = parser.read(nrows)
    455     finally:
    456         parser.close()

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\io\parsers.py in read(self, nrows)
   1131     def read(self, nrows=None):
   1132         nrows = _validate_integer("nrows", nrows)
-> 1133         ret = self._engine.read(nrows)
   1134 
   1135         # May alter columns / col_dict

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\io\parsers.py in read(self, nrows)
   2035     def read(self, nrows=None):
   2036         try:
-> 2037             data = self._reader.read(nrows)
   2038         except StopIteration:
   2039             if self._first_chunk:

pandas\_libs\parsers.pyx in pandas._libs.parsers.TextReader.read()

pandas\_libs\parsers.pyx in pandas._libs.parsers.TextReader._read_low_memory()

pandas\_libs\parsers.pyx in pandas._libs.parsers.TextReader._read_rows()

pandas\_libs\parsers.pyx in pandas._libs.parsers.TextReader._convert_column_data()

pandas\_libs\parsers.pyx in pandas._libs.parsers.TextReader._convert_tokens()

pandas\_libs\parsers.pyx in pandas._libs.parsers.TextReader._convert_with_dtype()

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\core\arrays\boolean.py in _from_sequence_of_strings(cls, strings, dtype, copy)
    302                 raise ValueError(f"{s} cannot be cast to bool")
    303 
--> 304         scalars = [map_string(x) for x in strings]
    305         return cls._from_sequence(scalars, dtype, copy)
    306 

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\core\arrays\boolean.py in <listcomp>(.0)
    302                 raise ValueError(f"{s} cannot be cast to bool")
    303 
--> 304         scalars = [map_string(x) for x in strings]
    305         return cls._from_sequence(scalars, dtype, copy)
    306 

c:\users\abbooda\appdata\local\continuum\miniconda3\envs\divi\lib\site-packages\pandas\core\arrays\boolean.py in map_string(s)
    300                 return False
    301             else:
--> 302                 raise ValueError(f"{s} cannot be cast to bool")
    303 
    304         scalars = [map_string(x) for x in strings]

ValueError: 1.0 cannot be cast to bool

Problem description

I would expect that 1.0, 0.0 and None can be cast to a nullable boolean. Using pd.read_csv() without dtype assignment assigns float64 dtype. Why does "boolean" not work?

Expected Output

print(df)
>>>   boolean_column  string_column
0            True              a
1           False              b
2            <NA>              c

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.7.6.final.0
python-bits : 64
OS : Windows
OS-release : 10
machine : AMD64
processor : Intel64 Family 6 Model 158 Stepping 10, GenuineIntel
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : None.None

pandas : 1.0.4
numpy : 1.18.1
pytz : 2020.1
dateutil : 2.8.1
pip : 20.1.1
setuptools : 47.3.0.post20200616
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.15.0
pandas_datareader: None
bs4 : None
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : None
matplotlib : 3.2.1
numexpr : None
odfpy : None
openpyxl : 3.0.3
pandas_gbq : None
pyarrow : None
pytables : None
pytest : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
xlsxwriter : None
numba : None

[TomAugspurger]

I think this just hasn't been implemented yet. Converting floats came up when BooleanArray was added, but IIRC we kept things simple at first.

To fix, we'll want to update BooleanArray._from_sequence_of_string.

cc @jorisvandenbossche.

[aauss]

Is this intended/desired to accept 1/1.0, 0/0.0 asbooleans? If so, I could update BooleanArray._from_sequence_of_string.

[aauss]

Hey @TomAugspurger and @jorisvandenbossche , do you think I should create a PR to enhance the BooleanArray._from_sequence_of_string function? I think it is worth it since it is not that uncommon to encode Booleans with 0 and 1 in a CSV and thus it would be nice to allow setting the nullable boolean for this column when reading the CSV.

[TomAugspurger]

This was discussed a bit in the original issue / PR implementing boolean array. You might want to read through it to get some context.

But IIRC the decision to not accept them at first was just for simplifying the initial implementation. I think we'd be find accepting them now.

[jorisvandenbossche]

Yes, I think at least 0's and 1's should be fine to accept (other numbers might require some extra discussion)