def forward_sample_demo():
"""Show estimation of probabilities by forward sampling"""
from Samplers import BNSampler
sampler = BNSampler(asia)
cancer_index = sampler.variable_index('Cancer')
xray_index = sampler.variable_index('XRay')
for size in 10,100,1000,10000:
print 'Sample size: ', size
for j in range(5):
sample = []
for i in xrange(size):
sample.append(sampler.forward_sample())
cancer_count, both_count = 0, 0
for inst in sample:
if inst[cancer_index] == 'absent':
cancer_count += 1
if inst[xray_index] == 'abnormal':
both_count += 1
print 'Estimate of P(Cancer=absent): ', cancer_count/float(size)
print 'Estimate of P(Cancer=absent,XRay=abnormal): ', both_count/float(size)
try:
print 'Estimate of P(XRay=abnormal|Cancer=absent): ', both_count/float(cancer_count)
except ZeroDivisionError:
print '0/0!'
print
def rejection_sample_demo():
"""Show estimation of probabilities by rejection sampling"""
from Samplers import BNSampler
sampler = BNSampler(asia)
sampler.condition({'XRay':'abnormal'})
print "Samples with 'None's are rejected"
for i in xrange(100):
print sampler.rejection_sample()