Error Downloading MNIST

[keepforever]

After using this code from the book/github:

from sklearn.datasets import fetch_mldata
mnist = fetch_mldata('MNIST original')
mnist

I get the following error

--------------------------------------------------------------
OSError                      Traceback (most recent call last)
<ipython-input-5-13fc10446cca> in <module>()
      1 
      2 from sklearn.datasets import fetch_mldata
----> 3 mnist = fetch_mldata('MNIST original')
      4 mnist

E:\brian\Anaconda3\lib\site-packages\sklearn\datasets\mldata.py in fetch_mldata(dataname, target_name, data_name, transpose_data, data_home)
    156     # load dataset matlab file
    157     with open(filename, 'rb') as matlab_file:
--> 158         matlab_dict = io.loadmat(matlab_file, struct_as_record=True)
    159 
    160     # -- extract data from matlab_dict

E:\brian\Anaconda3\lib\site-packages\scipy\io\matlab\mio.py in loadmat(file_name, mdict, appendmat, **kwargs)
    134     variable_names = kwargs.pop('variable_names', None)
    135     MR = mat_reader_factory(file_name, appendmat, **kwargs)
--> 136     matfile_dict = MR.get_variables(variable_names)
    137     if mdict is not None:
    138         mdict.update(matfile_dict)

E:\brian\Anaconda3\lib\site-packages\scipy\io\matlab\mio5.py in get_variables(self, variable_names)
    290                 continue
    291             try:
--> 292                 res = self.read_var_array(hdr, process)
    293             except MatReadError as err:
    294                 warnings.warn(

E:\brian\Anaconda3\lib\site-packages\scipy\io\matlab\mio5.py in read_var_array(self, header, process)
    250            `process`.
    251         '''
--> 252         return self._matrix_reader.array_from_header(header, process)
    253 
    254     def get_variables(self, variable_names=None):

scipy/io/matlab/mio5_utils.pyx in scipy.io.matlab.mio5_utils.VarReader5.array_from_header (scipy\io\matlab\mio5_utils.c:7087)()

scipy/io/matlab/mio5_utils.pyx in scipy.io.matlab.mio5_utils.VarReader5.array_from_header (scipy\io\matlab\mio5_utils.c:6206)()

scipy/io/matlab/mio5_utils.pyx in scipy.io.matlab.mio5_utils.VarReader5.read_real_complex (scipy\io\matlab\mio5_utils.c:7549)()

scipy/io/matlab/mio5_utils.pyx in scipy.io.matlab.mio5_utils.VarReader5.read_numeric (scipy\io\matlab\mio5_utils.c:4252)()

scipy/io/matlab/mio5_utils.pyx in scipy.io.matlab.mio5_utils.VarReader5.read_element (scipy\io\matlab\mio5_utils.c:3836)()

scipy/io/matlab/streams.pyx in scipy.io.matlab.streams.GenericStream.read_string (scipy\io\matlab\streams.c:2119)()

scipy/io/matlab/streams.pyx in scipy.io.matlab.streams.GenericStream.read_into (scipy\io\matlab\streams.c:1911)()

OSError: could not read bytes

[ageron]

I'm not entirely sure, but I'm guessing this is due to the fact that Python 3.6 on MacOSX does not come with SSL certificates, you must install the certifi package manually, as explained in this StackOverflow Answer.

[ageron]

Alternatively, this might be due to a corrupted download. Checkout this StackOverflow Answer.

[adamrmelnyk]

I ran into this as well. Data was corrupted. Just removed the mldata as suggested from the Stack Overflow post. Thanks!

[ageron]

Thanks for your feedback @adamrmelnyk , good to know that the fix still works.

[Doc-Scott]

This happened for me too, I think the data got corrupted the first time due to a bad internet connection. I removed it from the shell using:

rm ~/scikit_learn_data/mldata/mnist-original.mat

Which is described in StackOverflow post above, minus the rm, frustratingly it took me several different approaches to realise that. I'm using Linux but I think rm works in any pythonshell

[andsec]

Hello @ageron ,

How do you re-download the corrupted mnist-original.mat file after you have deleted it?

[nil12285]

copy https://github.com/amplab/datascience-sp14/raw/master/lab7/mldata/mnist-original.mat file into your scikit data home dir.

How to find where is your "scikit data home dir"?

from sklearn.datasets.base import get_data_home 
print (get_data_home())

[andsec]

Thank you @nil12285! It worked perfectly.

[ttwstnow]

thank u @nil12285, it works for me

[rohith28]

I used following code, It worked for me.
Getting data from a github repo

from six.moves import urllib
from sklearn.datasets import fetch_mldata

from scipy.io import loadmat
mnist_alternative_url = "https://github.com/amplab/datascience-sp14/raw/master/lab7/mldata/mnist-original.mat"
mnist_path = "./mnist-original.mat"
response = urllib.request.urlopen(mnist_alternative_url)
with open(mnist_path, "wb") as f:
    content = response.read()
    f.write(content)
mnist_raw = loadmat(mnist_path)
mnist = {
    "data": mnist_raw["data"].T,
    "target": mnist_raw["label"][0],
    "COL_NAMES": ["label", "data"],
    "DESCR": "mldata.org dataset: mnist-original",
}
print("Success!")

[ageron]

The fetch_mldata() function is now deprecated since Scikit-Learn 0.20. You should use fetch_openml() instead. I added more details in #301, so I'll close this issue.

[shobhit1893]

You this - this will work.
#301 (comment)