just make your targets 1D vector, e.g. via regressor = np.squeeze(regressor)
or regressor = regressor[:, 0] Cheers, On Tue, 20 Aug 2013, Dmitry Smirnov wrote: > Hi! > I'm trying to run some simple classification to get used to PyMVPA. > Specifically, I'm trying to classify 4 types of stimuli presented using > mask of 4501 voxels. > Here is the error and code. I'm new to python, so not much of experience > in figuring out error messages. Thanks in advance for all the suggestions! > ERROR: > Traceback (most recent call last): > � File "<stdin>", line 1, in <module> > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/learner.py", line > 239, in __call__ > ��� return super(Learner, self).__call__(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/node.py", line 84, > in __call__ > ��� result = self._call(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/measures/base.py", line > 472, in _call > ��� return super(CrossValidation, self)._call(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/measures/base.py", line > 301, in _call > ��� result = node(sds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/learner.py", line > 239, in __call__ > ��� return super(Learner, self).__call__(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/node.py", line 84, > in __call__ > ��� result = self._call(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/measures/base.py", line > 583, in _call > ��� res = measure(dstest) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/learner.py", line > 239, in __call__ > ��� return super(Learner, self).__call__(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/node.py", line 84, > in __call__ > ��� result = self._call(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/clfs/base.py", line 438, > in _call > ��� pred = self.predict(ds) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/clfs/base.py", line 48, > in wrap_samples > ��� return fx(obj, data, *args, **kwargs) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/clfs/base.py", line 409, > in predict > ��� result = self._predict(dataset) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/base/types.py", line 35, > in extract_samples > ��� return fx(obj, data.samples) > � File "/scratch/braindata/DSmirnov/PyMVPA/mvpa2/clfs/knn.py", line 191, > in _predict > ��� votes[labels[nn]] += 1 > TypeError: unhashable type: 'numpy.ndarray' > CODE: > from mvpa2.suite import * > from scipy import * > # samples: 1730x4501 matrix, rows=timepoints, columns=voxels > # regressor: 1730x1 vector, rows=labels of stimuli > # chunks: 1730x1 vector, rows=labels of independent runs (5 of them) > ds = dataset_wizard(samples, targets = regressor, chunks = selector) > clf = kNN(k=1, dfx=one_minus_correlation, voting='majority') > cvte = CrossValidation(clf, NFoldPartitioner()) > cv_results = cvte(ds) > BR, Dima -- Yaroslav O. Halchenko, Ph.D. http://neuro.debian.net http://www.pymvpa.org http://www.fail2ban.org Senior Research Associate, Psychological and Brain Sciences Dept. Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755 Phone: +1 (603) 646-9834 Fax: +1 (603) 646-1419 WWW: http://www.linkedin.com/in/yarik _______________________________________________ Pkg-ExpPsy-PyMVPA mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
