def makeModel(Data, lam0=0.1, lam1=0.1, **kwargs):
PriorArgs = dict(lam1=lam1, lam0=lam0)
PriorArgs.update(kwargs)
allocModel = bnpy.allocmodel.DPMixtureModel('VB', gamma0=1.0)
obsModel = bnpy.obsmodel.BernObsModel('VB', Data=Data, **PriorArgs)
model = bnpy.HModel(allocModel, obsModel)
return model
def makeModel(Data, ECovMat='diagcovdata', sF=1.0, **kwargs):
PriorArgs = dict(sF=sF, ECovMat=ECovMat)
PriorArgs.update(kwargs)
allocModel = bnpy.allocmodel.DPMixtureModel('VB', gamma0=1.0)
obsModel = bnpy.obsmodel.GaussObsModel('VB', Data=Data, **PriorArgs)
model = bnpy.HModel(allocModel, obsModel)
return model
def setUp(self, N=100, Data=None, likelihood='Gauss', **kwargs):
GaussSpec.update(kwargs)
DPSpec.update(kwargs)
if Data is None:
Data = makeGaussianDataWithTwoComps(N=N)
self.Data = Data
self.oneresp = 1e-50 * np.ones((2 * N, 2))
self.oneresp[:, 0] = 1.0
self.trueresp = 1e-50 * np.ones((2 * N, 2))
self.trueresp[:N, 0] = 1.0
self.trueresp[N:, 1] = 1.0
amodel = bnpy.allocmodel.DPMixtureModel('VB', DPSpec)
if likelihood.count('ZeroMean'):
omodel = bnpy.obsmodel.ZeroMeanGaussObsModel('VB',
Data=self.Data,
**GaussSpec)
else:
omodel = bnpy.obsmodel.GaussObsModel('VB',
Data=self.Data,
**GaussSpec)
self.hmodel = bnpy.HModel(amodel, omodel)