def setBinomialDistrbutionScore(self):
sizeOfDataSeries = self.genData.modelTargetData.size
freqDataSeries, binDataSeries = np.histogram(
self.genData.modelTargetData, bins=2)
probabilityOfTest_01, probabilityOfTest_02 = freqDataSeries / sizeOfDataSeries
expectedBinomialDistributionData_01 = stats.binom(
sizeOfDataSeries,
probabilityOfTest_01,
)
expectedBinomialDistributionData_02 = stats.binom(
sizeOfDataSeries, probabilityOfTest_02)
statistic_01, p_valueBinomial_01 = cvm_test(
self.genData.modelTargetData, expectedBinomialDistributionData_01)
statistic_01, p_valueBinomial_02 = cvm_test(
self.genData.modelTargetData, expectedBinomialDistributionData_02)
if p_valueBinomial_01 < 0.05 or p_valueBinomial_02 < 0.05:
self.DistributionScoreBinomial = {
'Binomial': (False, max(p_valueBinomial_01,
p_valueBinomial_02))
}
else:
self.DistributionScoreBinomial = {
'Binomial': (True, min(p_valueBinomial_01, p_valueBinomial_02))
}
def setPoissonDistributionScore(self):
mu = self.genData.modelTargetData.mean()
expectedPoissonDistributionData = stats.poisson(mu)
statistic, p_value = cvm_test(self.genData.modelTargetData,
expectedPoissonDistributionData)
if p_value < 0.05:
self.DistributionScorePoisson = {'Poisson': (False, p_value)}
else:
self.DistributionScorePoisson = {'Poisson': (True, p_value)}
def NZCVM(self): """ Compute the CramervonMises statistic and p-value for the two distributions of sumpz and true_z vector of spec-z's Parameters: ----------- using: string which parameterization to evaluate Returns: -------- CvM statistic and pvalue """ #copy the form of Rongpu's use of skgof functions #will have to use QPPDFCDF class, as those expect objects #that have a .cdf method for a vector of values tmpnzfunc = QPPDFCDF(self.stackpz) nzCvM = skgof.cvm_test(self.truth,tmpnzfunc) return nzCvM.statistic, nzCvM.pvalue
def CvM(self, using, dx=0.0001):
"""
Compute the Cramer-von Mises statistic and p-value for the PIT values
by comparing with a uniform distribution between 0 and 1.
Parameters:
-----------
using: string
which parameterization to evaluate
dx: float
step size for integral
Returns:
--------
CvM statistic and pvalue
"""
if self.pitarray is not None:
pits = np.array(self.pitarray)
else:
pits = np.array(self.PIT(using=using,dx=dx))
self.pitarray = pits
cvm_result = skgof.cvm_test(pits, stats.uniform())
return cvm_result.statistic, cvm_result.pvalue