Confuses X in GaussianHMM.fit ([X])

Question

With this code:

X = numpy.array(range(0,5)) model = GaussianHMM(n_components=3,covariance_type='full', n_iter=1000) model.fit([X]) 

I get

 tuple index out of range self.n_features = obs[0].shape[1] 

So what should you pass .fit () exactly? Hidden states and outliers in a tuple? If so, in what order? The documentation is less useful.

I noticed that he likes to go through tuples, as this does not give an error:

 X = numpy.column_stack([range(0,5),range(0,5)]) model = GaussianHMM(n_components=3,covariance_type='full', n_iter=1000) model.fit([X]) 

Edit:

Let me clarify a bit, the documentation indicates that the routine of the array should be:

List of observation sequences similar to an array (form (n_i, n_features)).

This almost means that you pass a tuple for each sample, which indicates in a binary way which observations are present. However, their example indicates otherwise:

 # pack diff and volume for training X = np.column_stack([diff, volume]) 

therefore confusion

Answer 1

It would seem that the Gaussian HMM function is intended for problems with multidimensional emission of only HMM, therefore, it is required to have> 1 emission vectors. When “n_features” is mentioned in the documentation, they are not among the methods of emission extraction, but rather the number of orthogonal emission vectors.

Therefore, “signs” (orthogonal emission vectors) should not be confused with “symbols”, which sklearn believes (which are likely to be shared with the large hmm community for everyone I know), refer to what actual unique values the system can radiate.

For one-dimensional emission vector problems, use MultinomialHMM.

Hope this is clarified by someone who wants to use this material without becoming the main authority in the world at HMM :)

Answer 2

I understand this is an old thread, but the problem in the sample code still exists. I believe the example is now in this link and still giving the same error:

 tuple index out of range self.n_features = obs[0].shape[1] 

Offensive line of code: model = GaussianHMM(n_components=5, covariance_type="diag", n_iter=1000).fit(X)

What should be: model = GaussianHMM(n_components=5, covariance_type="diag", n_iter=1000).fit([X])