fit_transform tries to query non-existent node "0"

[vhranger]

fit_transform tries to query non-existent node "0"

[ubalklen]

The workaround is to fit and predict separately.

[ubalklen]

Looking at the code, it seems to be related to the particular way the nodes are passed for prediction inside the fit_transform method in the node2vec implementation. Not sure why, but the predictions are performed using a np.arange. This range won't reflect the actual nodes in the graph, so an exception is raised.

The other embedders don't do that.

[VHRanger]

Thanks, I'll take a look this week.

[VHRanger]

should be fixed in latest release, please update and confirm

[ubalklen]

Still not working.

>pip show nodevectors

Name: nodevectors
Version: 0.1.23
Summary: Fast network node embeddings
Home-page: https://github.com/VHRanger/nodevectors/
Author: Matt Ranger
Author-email: UNKNOWN
License: MIT
Location: e:\users\user\miniconda3\envs\graph\lib\site-packages
Requires: scipy, networkx, csrgraph, gensim, pandas, numpy, scikit-learn, numba
Required-by:

>python
Python 3.7.9 (default, Aug 31 2020, 17:10:11) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> from nodevectors import Node2Vec
>>> import networkx as nx
>>> G = nx.Graph()       
>>> G.add_edge("1", "2") 
>>> n2v = Node2Vec(n_components=128)
>>> n2v.fit_transform(G)
Making walks... Done, T=2.98
Mapping Walk Names... Done, T=0.04
Training W2V... Done, T=0.05
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "E:\Users\user\miniconda3\envs\graph\lib\site-packages\nodevectors\node2vec.py", line 155, in fit_transform
    .apply(self.predict)
  File "E:\Users\user\AppData\Roaming\Python\Python37\site-packages\pandas\core\series.py", line 3848, in apply
    mapped = lib.map_infer(values, f, convert=convert_dtype)
  File "pandas\_libs\lib.pyx", line 2329, in pandas._libs.lib.map_infer
  File "E:\Users\user\miniconda3\envs\graph\lib\site-packages\nodevectors\node2vec.py", line 170, in predict
    return self.model.wv.__getitem__(node_name)
  File "E:\Users\user\miniconda3\envs\graph\lib\site-packages\gensim\models\keyedvectors.py", line 353, in __getitem__
    return self.get_vector(entities)
  File "E:\Users\user\miniconda3\envs\graph\lib\site-packages\gensim\models\keyedvectors.py", line 471, in get_vector
    return self.word_vec(word)
  File "E:\Users\user\miniconda3\envs\graph\lib\site-packages\gensim\models\keyedvectors.py", line 468, in word_vec
    raise KeyError("word '%s' not in vocabulary" % word)
KeyError: "word '0' not in vocabulary"

I can see there is still a np.arrange being passed to the predict function.

[IdanAlbilia]

Any new info about this problem? im having the exact same issue..

[IdanAlbilia]

or one might know a way around it?

[ashkspark]

Not working for me! Will do fit and predict.