apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-04-30 07:16:23 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

function predictive_quantiles added

Browse source
This commit is contained in:
Ricardo 2014-03-18 16:02:36 +00:00
parent ef31b5f1c9
commit b44fee93c4
1 changed files with 19 additions and 23 deletions
Download patch file Download diff file Expand all files Collapse all files

42
GPy/likelihoods/mixed_noise.py
View file

@ -11,7 +11,7 @@ import itertools
class MixedNoise(Likelihood): class MixedNoise(Likelihood):
def __init__(self, likelihoods_list, name='mixed_noise'): def __init__(self, likelihoods_list, name='mixed_noise'):
#NOTE at the moment this likelihood only works for using a list of gaussians
super(Likelihood, self).__init__(name=name) super(Likelihood, self).__init__(name=name)
self.add_parameters(*likelihoods_list) self.add_parameters(*likelihoods_list)
@ -38,35 +38,32 @@ class MixedNoise(Likelihood):
return np.array([dL_dKdiag[ind==i].sum() for i in range(len(self.likelihoods_list))]) return np.array([dL_dKdiag[ind==i].sum() for i in range(len(self.likelihoods_list))])
def predictive_values(self, mu, var, full_cov=False, Y_metadata=None): def predictive_values(self, mu, var, full_cov=False, Y_metadata=None):
if all([isinstance(l, Gaussian) for l in self.likelihoods_list]): ind = Y_metadata['output_index'].flatten()
ind = Y_metadata['output_index'].flatten() _variance = np.array([self.likelihoods_list[j].variance for j in ind ])
_variance = np.array([self.likelihoods_list[j].variance for j in ind ]) if full_cov:
if full_cov: var += np.eye(var.shape[0])*_variance
var += np.eye(var.shape[0])*_variance
else:
var += _variance
return mu, var
else: else:
raise NotImplementedError var += _variance
return mu, var
def predictive_variance(self, mu, sigma, **other_shit): def predictive_variance(self, mu, sigma, Y_metadata):
if isinstance(noise_index,int): _variance = self.gaussian_variance(Y_metadata)
_variance = self.variance[noise_index]
else:
_variance = np.array([ self.variance[j] for j in noise_index ])[:,None]
return _variance + sigma**2 return _variance + sigma**2
def predictive_quantiles(self, mu, var, quantiles, Y_metadata):
ind = Y_metadata['output_index'].flatten()
outputs = np.unique(ind)
Q = np.zeros( (mu.size,len(quantiles)) )
for j in outputs:
q = self.likelihoods_list[j].predictive_quantiles(mu[ind==j,:],
var[ind==j,:],quantiles,Y_metadata=None)
Q[ind==j,:] = np.hstack(q)
return [q[:,None] for q in Q.T]
def covariance_matrix(self, Y, Y_metadata): def covariance_matrix(self, Y, Y_metadata):
#assert all([isinstance(l, Gaussian) for l in self.likelihoods_list]) #TODO make more general, to allow non-gaussian likelihoods
#ind = Y_metadata['output_index'].flatten()
#variance = np.zeros(Y.shape[0])
#for lik, j in zip(self.likelihoods_list, range(len(self.likelihoods_list))):
# variance[ind==j] = lik.variance
#return np.diag(variance)
return np.diag(self.gaussian_variance(Y_metadata).flatten()) return np.diag(self.gaussian_variance(Y_metadata).flatten())
def samples(self, gp, Y_metadata): def samples(self, gp, Y_metadata):
""" """
Returns a set of samples of observations based on a given value of the latent variable. Returns a set of samples of observations based on a given value of the latent variable.
@ -84,4 +81,3 @@ class MixedNoise(Likelihood):
_ysim = np.array([np.random.normal(lik.gp_link.transf(gpj), scale=np.sqrt(lik.variance), size=1) for gpj in gp_filtered.flatten()]) _ysim = np.array([np.random.normal(lik.gp_link.transf(gpj), scale=np.sqrt(lik.variance), size=1) for gpj in gp_filtered.flatten()])
Ysim[flt,:] = _ysim.reshape(n1,N2) Ysim[flt,:] = _ysim.reshape(n1,N2)
return Ysim return Ysim