scikit-rebate Issue with Error message stating inputs are of wrong sizes

I am getting the following error, my inputs are of the correct input size. I am not sure where to contact the authors for assistance so writing this up as an issue. Xtrain is all ones and zeros (it comes from a 1hot encoder) and y is all ones and zeros. Example code given in documentation works fine so not an install issue.

I used the following code (I am running on a 16 core machine): fs = ReliefF(n_features_to_select=2000, n_neighbors=100,n_jobs=-1) fs.fit(Xtrain, y)

Shapes for additional info:

y.shape (3672,) Xtrain.shape (3672, 1688789)

Error message: Traceback (most recent call last): File "reliefChr19.py", line 180, in fs.fit(Xtrain, y) File "/usr/local/lib/python2.7/dist-packages/skrebate/relieff.py", line 154, in fit self.attr = self._get_attribute_info() #Examines each feature and applies discrete_threshold to determine variable type. File "/usr/local/lib/python2.7/dist-packages/skrebate/relieff.py", line 280, in _get_attribute_info zlen = len(np.unique(z)) File "/usr/local/lib/python2.7/dist-packages/numpy/lib/arraysetops.py", line 210, in unique return _unique1d(ar, return_index, return_inverse, return_counts) File "/usr/local/lib/python2.7/dist-packages/numpy/lib/arraysetops.py", line 279, in _unique1d flag = np.concatenate(([True], aux[1:] != aux[:-1])) ValueError: all the input arrays must have same number of dimensions

May 09 '18 13:05 swzCuroverse

were you able to figure out this issue?

May 22 '18 22:05 ryanurbs

No, I have not. However, I know have a set of data that we can use that is public so I can give it to you for debugging purposes. I will write up a script in the next few days if that is useful?

May 23 '18 14:05 swzCuroverse

A script for reproducing this issue is helpful. Thank you!

One questions, what is type(Xtrain) in your codes?

May 23 '18 14:05 weixuanfu

Yeah i'm just trying to figure out if there is a bug that needs to be fixed or if there might be a user error issue. Last night we just released a new version with a number of updates/fixes. So perhaps that alone might resolve your issue.

May 23 '18 14:05 ryanurbs

Ok, I will update the version and try again.

May 23 '18 15:05 swzCuroverse

Xtrain was a sparse that I made using the 1-hot encoding in sci-kit learn that I convert to dense using Xtrain = Xtrain.todense(),

May 23 '18 15:05 swzCuroverse

I updated and it seems to be chugging away with no error, so a good sign!

May 23 '18 15:05 swzCuroverse