read_csv with MultiIndex on columns doesn't respect level's dtypes

[duboism]

I want to store a dataframe with hierarchical column index in a CSV. The index contains both string and integer levels. If I'm correct this should be done like that:

import numpy as np
import pandas as pd
n = 10
p = 6
names = ['alphabetic', 'numeric']
levels = [['a', 'b'], [0, 1, 2]]
column_index = pd.MultiIndex.from_product(levels, names=names)
data = np.random.rand(n, p)
df = pd.DataFrame(data=data, columns=column_index)
df.to_csv('test.csv')
read_df = pd.io.parsers.read_csv('test.csv', header=[0, 1], index_col=0)
# This assertion fails
assert(np.all(df.columns == read_df.columns))

As mentioned, the assertion fails. The problem is that in read_df, the second level ('numeric') contains strings ('0', '1', '2') instead of numbers.

Versions:

>>> pd.util.print_versions.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 2.7.6.final.0
python-bits: 64
OS: Linux
OS-release: 3.13.0-68-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.17.0+git8-gcac4ad2
nose: 1.3.1
pip: 1.5.4
setuptools: 3.3
Cython: 0.20.1post0
numpy: 1.8.2
scipy: 0.13.3
statsmodels: 0.5.0
IPython: 1.2.1
sphinx: 1.2.2
patsy: 0.4.1
dateutil: 1.5
pytz: 2012c
blosc: None
bottleneck: None
tables: 3.1.1
numexpr: 2.4.3
matplotlib: 1.3.1
openpyxl: 2.3.0-b2
xlrd: 0.9.2
xlwt: 0.7.5
xlsxwriter: None
lxml: 3.3.3
bs4: 4.2.1
html5lib: 0.999
httplib2: 0.8
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None

[jreback]

this is a dupe of #9435. no easy way to do this ATM.

[duboism]

Are you sure it's a dupe? I have found #9435 when looking for a solution and I have tested the solution you proposed but it doesn't help. My problem is the dtype of the columns.

[jreback]

yes I am sure. Its a dupe because the dtype kw doesn't apply to an index column Index or MultiIndex.