Python Markov Transition Matrix handler
This library includes functionality that handles common tasks involved in using Markov transition matrices. For general information, see the webpage
Python 3
This class defines an object that handles Markov transition information. To add this module to a Python 3+ project, simply clone this repository into your project. The included init.py file should ensure that the class is accessible.
To initialize a new transition matrix, simply call Markov.TransitionMatrix(). To specify a filename for output, call Markov.TransitionMatrix(filename) (the default filename is "data").
The matrix can accept any form of transition where the state is either a string or an integer - anything that is a valid dictionary key in Python is accepted. Simply call matrix.add_transition(prev_state, next_state) where matrix is a TransitionMatrix object.
To normalize the data into a proper stochastic matrix, call matrix.normalize(). This function is called by default when saving the data to a file, but this call can be useful if one wishes to view normalized data at any point.
To save data, call matrix.save(). This will save the matrix of transition counts as filename.json in the current working directory, and the normalized stochastic matrix as filename_norm.json. By default, the function call will also return the transition count matrix; calling matrix.save(norm=True) will return the normalized matrix.
To load data, call matrix.load(filename). This will load an existing file containing transition counts into the object, and regenerate the normalized matrix. Be warned, this is additive: at any point, loading data into the matrix will combine with the existing data to produce a new matrix.
Once data is loaded, this object is also able to create a Markov chain based on the data. Simply call matrix.initialize_chain() to start a new Markov chain. Once this has been called, matrix.get_next_outcome() can be called repeatedly to return successive states in a Markov chain.