v0.0.2 push

examples, more comments, cleaned up code

examples/asia_bayes_net.py

@@ -0,0 +1,80 @@
+# Asia Bayes Net
+# Contact: Jacob Schreiber
+#          [email protected]
+
+'''
+The Asia Bayesian Network. See a description here:  
+http://www.norsys.com/tutorials/netica/secA/tut_A1.htm
+'''
+
+from pomegranate import *
+
+# Create the distributions
+asia = DiscreteDistribution({ 'True' : 0.5, 'False' : 0.5 })
+tuberculosis = ConditionalDiscreteDistribution({
+	'True' : DiscreteDistribution({ 'True' : 0.2, 'False' : 0.80 }),
+	'False' : DiscreteDistribution({ 'True' : 0.01, 'False' : 0.99 })
+	}, [asia])
+
+smoking = DiscreteDistribution({ 'True' : 0.5, 'False' : 0.5 })
+lung = ConditionalDiscreteDistribution({
+	'True' : DiscreteDistribution({ 'True' : 0.75, 'False' : 0.25 }),
+	'False' : DiscreteDistribution({ 'True' : 0.02, 'False' : 0.98 })
+	}, [smoking] )
+bronchitis = ConditionalDiscreteDistribution({
+	'True' : DiscreteDistribution({ 'True' : 0.92, 'False' : 0.08 }),
+	'False' : DiscreteDistribution({ 'True' : 0.03, 'False' : 0.97})
+	}, [smoking] )
+
+tuberculosis_or_cancer = ConditionalDiscreteDistribution({
+	'True' : { 'True' : DiscreteDistribution({ 'True' : 1.0, 'False' : 0.0 }),
+			   'False' : DiscreteDistribution({ 'True' : 1.0, 'False' : 0.0 }),
+			 },
+	'False' : { 'True' : DiscreteDistribution({ 'True' : 1.0, 'False' : 0.0 }),
+				'False' : DiscreteDistribution({ 'True' : 0.0, 'False' : 1.0 })
+			  }
+	}, [tuberculosis, lung] )
+
+xray = ConditionalDiscreteDistribution({
+	'True' : DiscreteDistribution({ 'True' : .885, 'False' : .115 }),
+	'False' : DiscreteDistribution({ 'True' : 0.04, 'False' : 0.96 })
+	}, [tuberculosis_or_cancer] )
+
+dyspnea = ConditionalDiscreteDistribution({
+	'True' : { 'True' : DiscreteDistribution({ 'True' : 0.96, 'False' : 0.04 }),
+			   'False' : DiscreteDistribution({ 'True' : 0.89, 'False' : 0.11 })
+			 },
+	'False' : { 'True' : DiscreteDistribution({ 'True' : 0.82, 'False' : 0.18 }),
+	            'False' : DiscreteDistribution({ 'True' : 0.4, 'False' : 0.6 })
+	          }
+	}, [tuberculosis_or_cancer, bronchitis])
+
+# Make the states. Note the name can be different than the name of the state
+# can be different than the name of the distribution
+s0 = State( asia, name="asia" )
+s1 = State( tuberculosis, name="tuberculosis" )
+s2 = State( smoking, name="smoker" )
+s3 = State( lung, name="cancer" )
+s4 = State( bronchitis, name="bronchitis" )
+s5 = State( tuberculosis_or_cancer, name="TvC" )
+s6 = State( xray, name="xray" )
+s7 = State( dyspnea, name='dyspnea' )
+
+# Create the Bayesian network
+network = BayesianNetwork( "asia" )
+network.add_states([ s0, s1, s2, s3, s4, s5, s6, s7 ])
+network.add_transition( s0, s1 )
+network.add_transition( s1, s5 )
+network.add_transition( s2, s3 )
+network.add_transition( s2, s4 )
+network.add_transition( s3, s5 )
+network.add_transition( s5, s6 )
+network.add_transition( s5, s7 )
+network.add_transition( s4, s7 )
+network.bake()
+
+print "Has tuberculosis, is not a smoker, 80-20 chance he has bronchitis"
+observations = { 'tuberculosis' : 'True', 'smoker' : 'False', 
+				 'bronchitis' : DiscreteDistribution({ 'True' : 0.8, 'False' : 0.2 }) }
+beliefs = map( str, network.forward_backward( observations ) )
+print "\n".join( "{}\t\t{}".format( state.name, belief ) for state, belief in zip( network.states, beliefs ) )

examples/fsm_test.py

@@ -0,0 +1,43 @@
+# FSM test
+# Contact: Jacob Schreiber
+#          [email protected]
+
+from pomegranate import *
+
+# Create the states in the same way as you would an HMM
+a = State( NormalDistribution( 5, 1 ), "a" )
+b = State( NormalDistribution( 23, 1 ), "b" )
+c = State( NormalDistribution( 100, 1 ), "c" )
+
+# Create a FiniteStateMachine object 
+model = FiniteStateMachine( "test" )
+
+# Add the states in the same way
+model.add_states( [a, b, c] )
+
+# Add the transitions in the same manner
+model.add_transition( model.start, a, 1.0 )
+model.add_transition( a, a, 0.33 )
+model.add_transition( a, b, 0.33 )
+model.add_transition( b, b, 0.5 )
+model.add_transition( b, a, 0.5 )
+model.add_transition( a, c, 0.33 )
+model.add_transition( c, a, 0.5 )
+model.add_transition( c, c, 0.5 )
+
+# Bake the model in the same way
+model.bake( verbose=True )
+
+# Take a sequence of observations
+seq = [ 5, 5, 5, 5, 23, 23, 5, 23, 23, 100, 23, 23, 23, 23, 5, 5, 100, 5, 23 ]
+
+# Print out the model 
+print "\n".join( state.name for state in model.states )
+
+# Print out where you start in the model
+print model.current_state.name
+
+# Print out where the model is for each step
+for symbol in seq:
+	model.step( symbol )
+	print symbol, model.current_state.name

examples/hmm_example.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python2.7
+# example.py: Yet Another Hidden Markov Model library
+# Contact: Jacob Schreiber ( [email protected] )
+
+"""
+A simple example highlighting how to build a model using states, add
+transitions, and then run the algorithms, including showing how training
+on a sequence improves the probability of the sequence.
+"""
+
+import random
+from pomegranate import *
+from pomegranate import HiddenMarkovModel as Model
+
+random.seed(0)
+model = Model(name="ExampleModel")
+distribution = UniformDistribution(0.0, 1.0)
+state = State(distribution, name="uniform")
+state2 = State(NormalDistribution(0, 2), name="normal")
+silent = State(None, name="silent")
+model.add_state(state)
+model.add_state(state2)
+
+model.add_transition(state, state, 0.4)
+model.add_transition(state, state2, 0.4)
+model.add_transition(state2, state2, 0.4)
+model.add_transition(state2, state, 0.4)
+
+model.add_transition(model.start, state, 0.5)
+model.add_transition(model.start, state2, 0.5)
+model.add_transition(state, model.end, 0.2)
+model.add_transition(state2, model.end, 0.2)
+
+model.bake()
+sequence = model.sample()
+print sequence
+print
+print model.forward(sequence)[ len(sequence), model.end_index ]
+print model.backward(sequence)[0,model.start_index]
+print
+trans, ems =  model.forward_backward(sequence)
+print trans
+print ems
+print
+model.train( [ sequence ] )
+
+print
+print model.forward(sequence)[ len(sequence), model.end_index ]
+print model.backward(sequence)[0,model.start_index]
+print
+trans, ems = model.forward_backward(sequence)
+print trans
+print ems
+print

examples/infinite_hmm.py

@@ -0,0 +1,85 @@
+# infinite_hmm_sampling.py
+# Contact: Jacob Schreiber
+#          [email protected]
+
+'''
+This example shows how to use yahmm to sample from an infinite HMM. The premise
+is that you have an HMM which does not have transitions to the end state, and
+so can continue on forever. This is done by not adding transitions to the end
+state. If you bake a model with no transitions to the end state, you get an
+infinite model, with no extra work! This change is passed on to all the
+algorithms.
+'''
+
+from pomegranate import *
+from pomegranate import HiddenMarkovModel as Model
+import itertools as it
+import numpy as np
+
+# Define the states
+s1 = State( NormalDistribution( 5, 2 ), name="S1" )
+s2 = State( NormalDistribution( 15, 2 ), name="S2" )
+s3 = State( NormalDistribution( 25, 2 ), name="S3 ")
+
+# Define the transitions
+model = Model( "infinite" )
+model.add_transition( model.start, s1, 0.7 )
+model.add_transition( model.start, s2, 0.2 )
+model.add_transition( model.start, s3, 0.1 )
+model.add_transition( s1, s1, 0.6 )
+model.add_transition( s1, s2, 0.1 )
+model.add_transition( s1, s3, 0.3 )
+model.add_transition( s2, s1, 0.4 )
+model.add_transition( s2, s2, 0.4 )
+model.add_transition( s2, s3, 0.2 )
+model.add_transition( s3, s1, 0.05 )
+model.add_transition( s3, s2, 0.15 )
+model.add_transition( s3, s3, 0.8 )
+model.bake()
+
+sequence = [ 4.8, 5.6, 24.1, 25.8, 14.3, 26.5, 15.9, 5.5, 5.1 ]
+
+
+print model.is_infinite()
+
+print "Algorithms On Infinite Model"
+sequence = [ 4.8, 5.6, 24.1, 25.8, 14.3, 26.5, 15.9, 5.5, 5.1 ]
+print "Forward"
+print model.forward( sequence )
+
+print "\n".join( state.name for state in model.states )
+print "Backward"
+print model.backward( sequence )
+
+print "Forward-Backward"
+trans, emissions = model.forward_backward( sequence )
+print trans
+print emissions
+
+print "Viterbi"
+prob, states = model.viterbi( sequence )
+print "Prob: {}".format( prob )
+print "\n".join( state[1].name for state in states )
+print
+print "MAP"
+prob, states = model.maximum_a_posteriori( sequence )
+print "Prob: {}".format( prob )
+print "\n".join( state[1].name for state in states )
+
+print "Showing that sampling can reproduce the original transition probs."
+print "Should produce a matrix close to the following: "
+print " [ [ 0.60, 0.10, 0.30 ] "
+print "   [ 0.40, 0.40, 0.20 ] "
+print "   [ 0.05, 0.15, 0.80 ] ] "
+print
+print "Tranition Matrix From 100000 Samples:"
+sample, path = model.sample( 100000, path=True )
+trans = np.zeros((3,3))
+
+for state, n_state in it.izip( path[1:-2], path[2:-1] ):
+	state_name = float( state.name[1:] )-1
+	n_state_name = float( n_state.name[1:] )-1
+	trans[ state_name, n_state_name ] += 1
+
+trans = (trans.T / trans.sum( axis=1 )).T
+print trans