def test_sourceprior_save_default(self):
prior = sourcePrior.SourcePrior("testSourceA")
prior.save()
retrievedPrior = sourcePrior.SourcePrior().get_prior(
source="testSourceA")
self.assertTrue(retrievedPrior.source == prior.source,
"values must be the same")
def test_sourceprior_save(self):
prior = sourcePrior.SourcePrior("testSourceA",
mean=10.0,
variance=100.0,
cov=0.5)
prior.save()
retrievedPrior = sourcePrior.SourcePrior().get_prior(
source="testSourceA")
self.assertTrue(retrievedPrior.cov == prior.cov,
"values must be the same")
def getObservedValueEstimate(self, dataPoints, source, property):
# get the source predicate prior
self = PropertyPrior().get_prior(source, property)
kbPrior = sourcePrior.SourcePrior().get_prior(source)
estimatedMeanVariance = None
n = len(dataPoints)
dataY = [y for (_, y) in dataPoints]
mlMean = statistics.mean(dataY)
assumedKnownVariance = math.pow(kbPrior.cov * mlMean, 2)
# if this is the first time predicate is being observed in the source,
# then use the source prior as the prior variance for the source-predicate variance,
# and then use the mean of the observed values as the mean of the prior mean.
# update the source-predicate prior
# else if the predicate already has a source-property prior entry,
# then use the Gaussian params to estimate posterior mean and variance
# if spp != None:
estimatedMeanVariance = self.posterior(dataPoints,
assumedKnownVariance)
# update the source priors
kbPrior.posterior(dataPoints,
knownMean=estimatedMeanVariance[0],
knownVariance=estimatedMeanVariance[1])
return estimatedMeanVariance
def test_get_source_posterior(self):
prior = sourcePrior.SourcePrior("testSourceA",
mean=10.0,
variance=100.0,
cov=0.5)
data = [("A", 20.0), ("B", 25), ("C", 20), ("D", 30.0)]
posterior = prior.posterior(data, 20.0, 100.0)
self.assertTrue(posterior[0] > 0, "positive posterior required")
def test_get_prop_posterior(self):
kb_prior = sourcePrior.SourcePrior("testSourceA",
mean=10.0,
variance=100.0)
kb_prior.save()
prior = propertyPrior.PropertyPrior("testSourceA",
"propA",
mean=5.0,
variance=10)
data = [("A", 20.0), ("B", 25), ("C", 20), ("D", 30.0)]
posterior = prior.posterior(data, knownVariance=100.0)
self.assertTrue(posterior[0] > 0, "positive posterior required")
def test_get_observed_estimate_value(self):
kb_prior = sourcePrior.SourcePrior("testSourceA",
mean=10.0,
variance=100.0)
kb_prior.save()
prior = propertyPrior.PropertyPrior("testSourceA",
"propA",
mean=5.0,
variance=10)
data = [("A", 20.0), ("B", 25), ("C", 20), ("D", 30.0)]
posterior = prior.getObservedValueEstimate(data, "testSourceA",
"propA")
self.assertTrue(posterior[0] > 0, "positive posterior required")