Error when subtracting mixed type DataFrame

[aman-thakral]

Sample code to reproduce the error:

import numpy as np
import pandas

X = np.random.rand(10,10)
Y = np.ones((10,1),dtype=int)
df1 = pandas.DataFrame(X)
df1 = df1.join(pandas.DataFrame(Y),lsuffix='.X')

print df1-df1.mean()

---

Traceback (most recent call last):
File "pls_model.py", line 383, in
run(df)
File "pls_model.py", line 305, in run
report(df.copy(),n_comp,priors=priors,plot=False)
File "pls_model.py", line 83, in report
plsgnb.fit(X,Y,priors=priors)
File "C:\src\multivariate\src\multivariate\PLS.py", line 477, in fit
self.model.fit(X,Y)
File "C:\src\multivariate\src\multivariate\PLS.py", line 145, in fit
print x - self.Xmean
File "C:\src\pandas\pandas\core\frame.py", line 177, in f
return self._combine_series(other, func, fill_value, axis, level)
File "C:\src\pandas\pandas\core\frame.py", line 2526, in _combine_series
return self._combine_series_infer(other, func, fill_value)
File "C:\src\pandas\pandas\core\frame.py", line 2541, in _combine_series_infer
return self._combine_match_columns(other, func, fill_value)
File "C:\src\pandas\pandas\core\frame.py", line 2552, in _combine_match_columns
left, right = self.align(other, join='outer', axis=1, copy=False)
File "C:\src\pandas\pandas\core\frame.py", line 1735, in align
method=method)
File "C:\src\pandas\pandas\core\frame.py", line 1804, in _align_series
return (left_result.fillna(fill_value, method=method),
File "C:\src\pandas\pandas\core\frame.py", line 2426, in fillna
new_data = self._data.fillna(value)
File "C:\src\pandas\pandas\core\internals.py", line 938, in fillna
new_blocks = [b.fillna(value) for b in self.blocks]
File "C:\src\pandas\pandas\core\internals.py", line 211, in fillna
new_values.flat[mask] = value
ValueError: cannot convert float NaN to integer

[aman-thakral]

I just checked this against 0.7.0 and the problem is not present. This is related to 0.7.1. Would this problem also be a good candidate to include in the tests?

[adamklein]

Absolutely, needs to get in the test suite and fixed! Good catch

[wesm]

hi @aman-thakral fixed a pair of related issue around this. thanks for the report