def test_constructor_with_params():
# Custom Parameters
tempDic = {
'dem': (.6, .4),
'target': (0.06, ),
'feat': ('mean_only_mvn', [[4, 2], [2, 4]]),
'featnoise': ('shape_spread_only_mvn', 0.9)
}
testPopulation = bias.Population(parameter_dictionary=tempDic)
# Assigns Custom parameters to individual components
testDem = bias.Demographic(.6, .4)
testTarget = bias.Target(0.06)
testFeature = bias.Feature('mean_only_mvn', [[4, 2], [2, 4]])
testFeatureNoise = bias.FeatureNoise('shape_spread_only_mvn', 0.9)
# Creates description of the parameters the same way Populations does
description = ''
description += 'Demographic Parameters\n'
description += testDem.params.__str__()
description += '\nTarget Parameters \n'
description += testTarget.params.__str__()
description += '\nFeature Parameters \n'
description += testFeature.params.__str__()
description += '\nFeature Noise Parameters \n'
description += testFeatureNoise.params.__str__()
# Compares parameters to make sure they are set correctly in Populations
assert testPopulation.get_parameter_description() == description
def test_target_sampler():
testPop = bias.Population()
# Number of Samples used in Population class
sampleNum = 4000
# How close to the target parameter it has to be
accuracyThreshold = .01
df = testPop.sample(sampleNum)
probY = sum(df['y']) / sampleNum
probZ = sum(df['z']) / sampleNum
# Checks to make sure target probability and actual are close enough
assert abs(probY - probZ) < accuracyThreshold
def test_base_constructor():
test_pop = bias.Population()
# Makes sure the default parameters are set correctly
assert test_pop.demographic_sampler.__dict__ == bias.bias_components.Demographic(
).__dict__
assert test_pop.feature_sampler.__dict__ == bias.bias_components.Feature(
).__dict__
assert test_pop.target_sampler.__dict__ == bias.bias_components.Target(
).__dict__
assert test_pop.feature_noise_sampler.__dict__ == bias.bias_components.FeatureNoise(
).__dict__
def test_demographic_sampler():
testPop = bias.Population()
#Number of samples used in Population class
sampleNum = 3000
# How close to the target parameter it has to be
accuracyThreshold = .03
df = testPop.sample(sampleNum)
probA = sum(df['a']) / sampleNum
probZ = sum(df['z']) / sampleNum
# Checks to make sure target probability and actual are close enough
assert abs(probA -
testPop.demographic_sampler.get_rho_a()) < accuracyThreshold
assert abs(probZ -
testPop.demographic_sampler.get_rho_z()[0]) < accuracyThreshold
def test_overall():
assert (bias.Population())