I am trying to train a Hidden Markov Model using hmmlearn. I have 4 features, out of which 3 are continuous and follow a Gaussian probability distribution. My 4th feature, however, is discrete and only takes the values 1 or 2. I am trying to classify my data into 2 classes (ie, n_components = 2). How can I train a model that handles these features correctly?
Any insight is much appreciated. Thanks.
I tried using GaussianHMM with all 4 features as the input. This did not give me totally bad results but I would like to make sure I am handling all the features correctly.