Something is wrong in the ClusterGenerator

[posmat]

There is a line:
if not all(hasattr(elem, 'iter') and len(elem) == self.n_feats for elem in self.distributions)
raise ValueError('Invalid distributions input! Input must have dimensions (n_clusters, n_feats).')

Because self.distributions is a list, then len(elem) is the character number of the distribution string. e.g. 'gaussian' is 8 character, so this line demands that the n_feats to be 8

[dcferreira]

Hi @orbennix, thanks for the message.

I haven't touched this code in a while, so forgive me if I say something wrong.
As far as I can tell, you're referring to line 149 here, right?

mdcgenpy/mdcgenpy/clusters/__init__.py

Lines 144 to 152 in 38f73da

	# check validity of self.distributions, and turning it into a (n_clusters, n_feats) matrix
	if hasattr(self.distributions, '__iter__') and not type(self.distributions) == str:
	if len(self.distributions) != self.n_clusters:
	raise ValueError('There must be exactly one distribution input for each cluster!')
	if hasattr(self.distributions[0], '__iter__'):
	if not all(hasattr(elem, '__iter__') and len(elem) == self.n_feats for elem in self.distributions):
	raise ValueError('Invalid distributions input! Input must have dimensions (n_clusters, n_feats).')
	else:
	self.distributions = [self.distributions] * self.n_clusters

As far as I can tell, this line checks if the input for the distributions is a matrix of dimension n_clusters x n_feats.
Indeed if you send it a list of strings (as opposed to a list of lists of strings), the check would verify the wrong thing.

Would you care to make a pull request for this issue?