Orange Forum • View topic - Hidden Markov Models

Hidden Markov Models

A place to ask questions about methods in Orange and how they are used and other general support.

Hidden Markov Models

Postby roschler » Thu Apr 28, 2005 8:25

WARNING: I failed statistics! :roll:

Is there anything in Orange that can provide the same functionality as a Hidden Markov Model does?

I'm looking for something that can be trained on data events and be used to predict the probability of the next event based on the previous events.

I just printed out the Introductory Orange paper and I'll be reading it tomorrow.

If Orange can't do this (yet?), can you point me to something that does?

Thanks.

Postby Blaz » Thu Apr 28, 2005 14:08

orange does not include hidden markov models. if this is the technique you want to use, there should be plenty implementations available. alernatively, you may use machine learning for similar tasks. to do this, construct a data table where each row consist of attributes that encode occurence of events in time t-1, t-2, ..., t-n (n attributes) and a class which encodes the occurence of event at time t. this provides ground for learning a model for probability of event at some time t based on n previous occurecies. a machine learning method of your choice can be used on such data set, and the problem becomes a typical ML problem.


Return to Questions & Support