pandas ValueError: numpy.dtype has the wrong size,

Question

I took a new clean install of OSX 10.9.3 and installed pip, and then did

pip install pandas
pip install numpy

Both installs seemed to be perfectly happy, and ran without any errors (though there were a zillion warnings). When I tried to run a python script with import pandas, I got the following error:

    numpy.dtype has the wrong size, try recompiling Traceback (most recent call last): 
    File "./moen.py", line 7, in  import pandas File "/Library/Python/2.7/site-packages/pandas/__init__.py", line 6, in  from . import hashtable, tslib, lib 
    File "numpy.pxd", line 157, in init pandas.hashtable (pandas/hashtable.c:22331) 
    ValueError: numpy.dtype has the wrong size, try recompiling

How do I fix this error and get pandas to load properly?

Answer 1

sudo pip install pandas
sudo easy_install --upgrade numpy

should also realign everything.

Answer 2

You can install previous version of pandas.

pip uninstall numpy
pip uninstall pandas
pip install pandas==0.13.1

In my situation it solved problem...

Answer 3

Uninstall both numpy and pandas and try installing pandas from source.

pip uninstall numpy
pip uninstall pandas
git clone git://github.com/pydata/pandas.git
cd pandas
python setup.py install

This worked for me and I am now able to use the latest version of pandas.

Answer 4

pip uninstall numpy uninstalls the old version of numpy

pip install numpy finds and installs the latest version of numpy

Answer 5

open your python, check the imported version of your numpy.

It is very likely that you have multiple numpy installed and python always grab the old one, just make sure to delete the old one would fix the problem.

>>> import numpy as np
>>> np.__version__
>>> np.__file__
#if numpy version <= 1.7 would have the error
#find the file and delete it from (np.__file__)

then install the latest numpy if you don't have it

Answer 6

you can install pandas from its git repo without having to explicitly clone it

pip install git+https://github.com/pydata/pandas.git

that worked for me.