Scikit-learn - Cannot load MNIST Original dataset using fetch_openml in Python

Question

I'm trying to load the MNIST Original dataset in Python. The sklearn.datasets.fetch_openml function doesn't seem to work for this.

Here is the code I'm using-

from sklearn.datasets import fetch_openml
dataset = fetch_openml("MNIST Original")

I get this error-

File "generateClassifier.py", line 11, in <module>
  dataset = fetch_openml("MNIST Original")
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 526, in fetch_openml
data_info = _get_data_info_by_name(name, version, data_home)
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 302, in 
_get_data_info_by_name
    data_home)
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 169, in 
_get_json_content_from_openml_api
    raise error
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 164, in 
_get_json_content_from_openml_api
    return _load_json()
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 52, in wrapper
    return f()
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 160, in _load_json
    with closing(_open_openml_url(url, data_home)) as response:
  File "/home/inglorion/.local/lib/python3.5/site- 
packages/sklearn/datasets/openml.py", line 109, in _open_openml_url
with closing(urlopen(req)) as fsrc:
  File "/usr/lib/python3.5/urllib/request.py", line 163, in urlopen
    return opener.open(url, data, timeout)
  File "/usr/lib/python3.5/urllib/request.py", line 472, in open
    response = meth(req, response)
  File "/usr/lib/python3.5/urllib/request.py", line 582, in 
http_response
    'http', request, response, code, msg, hdrs)
  File "/usr/lib/python3.5/urllib/request.py", line 510, in error
    return self._call_chain(*args)
  File "/usr/lib/python3.5/urllib/request.py", line 444, in 
_call_chain
    result = func(*args)
  File "/usr/lib/python3.5/urllib/request.py", line 590, in 
http_error_default
    raise HTTPError(req.full_url, code, msg, hdrs, fp)
        urllib.error.HTTPError: HTTP Error 400: Bad Request

How can I fix this? Alternately, is there any other way to load the MNIST dataset into Python?

I'm using version 0.20.2 of scikit-learn.

I'm relatively new to programming in general, so I would appreciate it if I could get a simple answer. Thanks!

That didn't work for me either; I think fetch_mldata is soon to be deprecated anyway. — Inglorion, Jan 25 '19 at 12:17
checkout the answers on this thread https://stackoverflow.com/questions/47324921/cant-load-mnist-original-dataset-using-sklearn — DrBwts, Jan 25 '19 at 12:25
@DrBrwts : I did, it seems to be a problem unrelated to what I'm facing, as I got a totally different error... — Inglorion, Jan 25 '19 at 12:28
from sklearn import datasets data = datasets.load_digits() I hope that's enough of the original MNIST data for you — charley, Jan 25 '19 at 12:38

score 21 · Answer 1 · answered Mar 05 '19 at 15:40

21

Try

mnist = fetch_openml('mnist_784')

I found it via https://www.openml.org/ under https://www.openml.org/d/554

answered Mar 05 '19 at 15:40

bav av

211
1
4

daz · Answer 2 · 2021-05-23T05:43:38.323

5

Method fetch_openml() download dataset from mldata.org which is not stable and can not connect. An alternative way is manually to download the data set from the original data. You can download data from Kaggle(mnist data) and run the following code

from scipy.io import loadmat
mnist = loadmat("../input/mnist-original.loadmat")
mnist_data = mnist["data"].T
mnist_label = mnist["label"][0]

edited May 23 '21 at 05:43

answered May 23 '21 at 05:33

daz

696
10
10

score 3 · Answer 3 · edited Mar 22 '20 at 14:50

3

you can use:

mist = fetch_openml('mnist_784', version=1)

edited Mar 22 '20 at 14:50

Mark

934
1
10
25

answered Mar 22 '20 at 12:21

bforblack

65
5

score 2 · Answer 4 · answered Feb 28 '20 at 10:41

fetch_mldata is deprecated since scikit-learn v0.20

Test sklearn version

import sklearn
sklearn.__version__

Import Dataset

from sklearn.datasets import fetch_openml
X, y = fetch_openml('mnist_784', version=1, return_X_y=True)

Example

https://scikit-learn.org/stable/auto_examples/neural_networks/plot_mnist_filters.html

score 0 · Answer 5 · answered Apr 05 '21 at 01:05

I was also facing similar problem. Updating sklearn's version worked for me

I just ran the following command

conda update scikit-learn

Then to verify the version, you can do something like this

import nltk
import sklearn

print('nltk version: {}.'.format(nltk.__version__))
print('scikit-learn version: {}.'.format(sklearn.__version__))

Do not forget to restart the kernel after updating the sklearn's version.

Scikit-learn - Cannot load MNIST Original dataset using fetch_openml in Python

5 Answers5

Test sklearn version

Import Dataset

Example