def generate_negative_ground_truth_predicted_targets():
ground_truth = -np.ones(4)
predicted_targets = np.array([-1., -2., -3., -4.])
distributions = [
distr.ReparametrizedGaussian(y, 0.) for y in predicted_targets
]
predictiveDistribution = distr.PredictiveDistribution(distributions)
return ground_truth, predictiveDistribution
def generate_identical_ground_truth_predicted_targets():
ground_truth = np.array([-1., -2., -3., -4.])
predicted_targets = ground_truth
distributions = [
distr.ReparametrizedGaussian(y, 0.) for y in predicted_targets
]
predictiveDistribution = distr.PredictiveDistribution(distributions)
return ground_truth, predictiveDistribution
def generate_random_labels_and_predictions_and_predictive_distribution():
np.random.seed(1)
ground_truth = np.random.rand(30)
predicted_targets = np.random.rand(30)
distributions = [
distr.ReparametrizedGaussian(target, 0.)
for target in predicted_targets
]
predictiveDistribution = distr.PredictiveDistribution(distributions)
return ground_truth, predicted_targets, predictiveDistribution