py-mm

Markov models for categorical sequential data in Python. The joint probabiity

for an ordered sequence of the categorical variable is composed of M+1 conditional distributions of order . Each component is represented by a rank m+1 tensor of transition state probabilities. These are inferred under maximum likelihood estimation of the data.

import numpy as np
from pymm.models import MarkovModel


# number of states for categorical variable
K = 2

# order of Markov model
M = 1

# generator for synthetic data
generator = MarkovModel(K=K, M=M, random_init=True)

# create artificial sequential dataset
X = [generator.sample(10) for _ in range(1000)]

# Markov model to infer joint distribution for sequential data
model = MarkovModel(K=K, M=M)

# infer conditional transition state probabilities
%timeit -r 1 -n 1 model.fit(X)

Install

To install, download or git clone the full repository and then run

python setup.py install

from the repository root in your chosen python environment. PyPi release to follow.

Status

This repository is pre-release and in active development. Please check for updates and switch to the PyPi release when available.