1# Copyright (c) 2012 - 2014 the GPy Austhors (see AUTHORS.txt) 2# Licensed under the BSD 3-clause license (see LICENSE.txt) 3 4import numpy as np 5from ..core import GP 6from .. import likelihoods 7from .. import kern 8 9class GPRegression(GP): 10 """ 11 Gaussian Process model for regression 12 13 This is a thin wrapper around the models.GP class, with a set of sensible defaults 14 15 :param X: input observations 16 :param Y: observed values 17 :param kernel: a GPy kernel, defaults to rbf 18 :param Norm normalizer: [False] 19 :param noise_var: the noise variance for Gaussian likelhood, defaults to 1. 20 21 Normalize Y with the norm given. 22 If normalizer is False, no normalization will be done 23 If it is None, we use GaussianNorm(alization) 24 25 .. Note:: Multiple independent outputs are allowed using columns of Y 26 27 """ 28 29 def __init__(self, X, Y, kernel=None, Y_metadata=None, normalizer=None, noise_var=1., mean_function=None): 30 31 if kernel is None: 32 kernel = kern.RBF(X.shape[1]) 33 34 likelihood = likelihoods.Gaussian(variance=noise_var) 35 36 super(GPRegression, self).__init__(X, Y, kernel, likelihood, name='GP regression', Y_metadata=Y_metadata, normalizer=normalizer, mean_function=mean_function) 37 38 @staticmethod 39 def from_gp(gp): 40 from copy import deepcopy 41 gp = deepcopy(gp) 42 return GPRegression(gp.X, gp.Y, gp.kern, gp.Y_metadata, gp.normalizer, gp.likelihood.variance.values, gp.mean_function) 43 44 def to_dict(self, save_data=True): 45 model_dict = super(GPRegression,self).to_dict(save_data) 46 model_dict["class"] = "GPy.models.GPRegression" 47 return model_dict 48 49 @staticmethod 50 def _from_dict(input_dict, data=None): 51 import GPy 52 input_dict["class"] = "GPy.core.GP" 53 m = GPy.core.GP.from_dict(input_dict, data) 54 return GPRegression.from_gp(m) 55 56 def save_model(self, output_filename, compress=True, save_data=True): 57 self._save_model(output_filename, compress=True, save_data=True) 58