diff --git a/examples/analytical-function/example_analytical_function.py b/examples/analytical-function/example_analytical_function.py
index 70124f9cf8b4681c02ec99141d0d5c2fc78df11d..30ff3fac4dfd3390135aed838295742d7fcef932 100644
--- a/examples/analytical-function/example_analytical_function.py
+++ b/examples/analytical-function/example_analytical_function.py
@@ -237,8 +237,9 @@ if __name__ == "__main__":
PostPCE.check_accuracy(n_samples=300)
# Compute the moments and compare with the Monte-Carlo reference
- if MetaModelOpts.meta_model_type != 'GPE':
- PostPCE.plot_moments()
+ # TODO: generalize the moment calculation
+ #if MetaModelOpts.meta_model_type != 'GPE':
+ # PostPCE.plot_moments()
# Plot the evolution of the KLD,BME, and Modified LOOCV error
if MetaModelOpts.ExpDesign.method == 'sequential':
@@ -246,6 +247,7 @@ if __name__ == "__main__":
PostPCE.plot_seq_design_diagnostics(refBME_KLD)
# Plot the sobol indices
+ # TODO: generalize this by using moment calculation(?)
if MetaModelOpts.meta_model_type != 'GPE':
total_sobol = PostPCE.sobol_indices()
diff --git a/examples/user_guide/example_user_guide.py b/examples/user_guide/example_user_guide.py
index 4c191b7de270ef6a817bbf2fd3cc24b9d8d3f3a3..5a32c15d6d49493de221e02557e950036918b034 100644
--- a/examples/user_guide/example_user_guide.py
+++ b/examples/user_guide/example_user_guide.py
@@ -66,7 +66,7 @@ if __name__ == '__main__':
MetaMod.pce_q_norm = 1
# TODO: add check in Metamod/ExpDesign to ensure that n_init_samples matches the length of the given .X!
- ExpDesign.n_init_samples = 30
+ ExpDesign.n_init_samples = 40
ExpDesign.sampling_method = 'random'#'user'
#ExpDesign.X = samples
diff --git a/src/bayesvalidrox/surrogate_models/engine.py b/src/bayesvalidrox/surrogate_models/engine.py
index 8d27310b70737a05f207314b30af0e29be280001..92e5dbf663973ba74cfbbea7bb7411f8c2605eb2 100644
--- a/src/bayesvalidrox/surrogate_models/engine.py
+++ b/src/bayesvalidrox/surrogate_models/engine.py
@@ -263,9 +263,13 @@ class Engine:
pce = False
#mc_ref = True if bool(self.Model.mc_reference) else False
mc_ref = False
- if self.Model.mc_reference != {}:
+ # TODO: somehow get issues from these comparisons, thus doubled it here
+ if self.Model.mc_reference is not {}:
mc_ref = True
self.Model.read_observation('mc_ref')
+ #elif self.Model.mc_reference != {}:
+ # mc_ref = True
+ # self.Model.read_observation('mc_ref')
# Get the parameters
max_n_samples = self.ExpDesign.n_max_samples
@@ -348,7 +352,7 @@ class Engine:
for out_name in output_name:
y = self.ExpDesign.Y[out_name]
Scores_all.append(list(
- self.MetaModel.score_dict['b_1'][out_name].values()))
+ self.MetaModel.LOOCV_score_dict['b_1'][out_name].values()))
if self.MetaModel.dim_red_method.lower() == 'pca':
pca = self.MetaModel.pca['b_1'][out_name]
components = pca.transform(y)
@@ -494,7 +498,7 @@ class Engine:
for out_name in output_name:
y = self.ExpDesign.Y[out_name]
Scores_all.append(list(
- self.MetaModel.score_dict['b_1'][out_name].values()))
+ self.MetaModel.LOOCV_score_dict['b_1'][out_name].values()))
if self.MetaModel.dim_red_method.lower() == 'pca':
pca = self.MetaModel.pca['b_1'][out_name]
components = pca.transform(y)
@@ -1013,6 +1017,8 @@ class Engine:
"""
if self.Model.mc_reference is {}:
raise AttributeError('Model.mc_reference needs to be given to calculate the surrogate error!')
+ #elif self.Model.mc_reference == {}:
+ # raise AttributeError('Model.mc_reference needs to be given to calculate the surrogate error!')
# Compute the mean and std based on the MetaModel
means, stds = self.MetaModel.compute_moments()
diff --git a/src/bayesvalidrox/surrogate_models/polynomial_chaos.py b/src/bayesvalidrox/surrogate_models/polynomial_chaos.py
index fc483d57d4e324536ac9d69f29cfc4a9533bb166..103edfa8dae25b287bbd274e9a81be7c0aff2590 100644
--- a/src/bayesvalidrox/surrogate_models/polynomial_chaos.py
+++ b/src/bayesvalidrox/surrogate_models/polynomial_chaos.py
@@ -100,10 +100,10 @@ class PCE(MetaModel):
pce_reg_method='OLS', bootstrap_method='fast',
n_bootstrap_itrs=1, pce_deg=1, pce_q_norm=1.0,
dim_red_method='no', verbose=False):
- # Check if the result would be gaussian
-
+ # Check if the surrogate outputs gaussian results
is_gaussian = self.check_is_gaussian(pce_reg_method, n_bootstrap_itrs)
+ # Use parent init
super().__init__(input_obj, bootstrap_method,
n_bootstrap_itrs, dim_red_method, is_gaussian,
verbose)
@@ -121,22 +121,14 @@ class PCE(MetaModel):
self._error_clf_poly = None
self._errorRegMethod = None
self._univ_p_val = None
- #self.out_names = None
self._allBasisIndices = None
self._deg_array = None
- #self.n_samples = None
- #self.CollocationPoints = None # TODO: rename this!
- #self.pca = None
self._LCerror = None
self._clf_poly = None
- self._score_dict = None
self._basis_dict = None
self._coeffs_dict = None
self._q_norm_dict = None
self._deg_dict = None
- #self.n_params = None
- #self.ndim = None
- #self.rmse = None
def check_is_gaussian(self, _pce_reg_method, n_bootstrap_itrs) -> bool:
"""
@@ -202,7 +194,7 @@ class PCE(MetaModel):
self._q_norm_dict = self.auto_vivification()
self._coeffs_dict = self.auto_vivification()
self._basis_dict = self.auto_vivification()
- self._score_dict = self.auto_vivification()
+ self.LOOCV_score_dict = self.auto_vivification()
self._clf_poly = self.auto_vivification()
self._LCerror = self.auto_vivification()
if self.dim_red_method.lower() == 'pca':
@@ -376,7 +368,7 @@ class PCE(MetaModel):
self._q_norm_dict[f'b_{b_i + 1}'][key][f"y_{i + 1}"] = out[i]['qnorm']
self._coeffs_dict[f'b_{b_i + 1}'][key][f"y_{i + 1}"] = out[i]['coeffs']
self._basis_dict[f'b_{b_i + 1}'][key][f"y_{i + 1}"] = out[i]['multi_indices']
- self._score_dict[f'b_{b_i + 1}'][key][f"y_{i + 1}"] = out[i]['LOOCVScore']
+ self.LOOCV_score_dict[f'b_{b_i + 1}'][key][f"y_{i + 1}"] = out[i]['LOOCVScore']
self._clf_poly[f'b_{b_i + 1}'][key][f"y_{i + 1}"] = out[i]['_clf_poly']
# self._LCerror[f'b_{b_i+1}'][key][f"y_{i+1}"] = out[i]['_LCerror']
@@ -567,8 +559,8 @@ class PCE(MetaModel):
elif reg_method.lower() == 'ebl':
_clf_poly = EBLinearRegression(optimizer='em')
- print(X.shape)
- print(y.shape)
+ #print(X.shape)
+ #print(y.shape)
_clf_poly.fit(X, y)
# Select the nonzero entries of coefficients
@@ -576,6 +568,7 @@ class PCE(MetaModel):
nnz_idx = np.nonzero(_clf_poly.coef_)[0]
else:
nnz_idx = np.arange(_clf_poly.coef_.shape[0])
+ #print(nnz_idx)
# This is for the case where all outputs are zero, thereby
# all coefficients are zero
@@ -660,6 +653,8 @@ class PCE(MetaModel):
# Assemble the Psi matrix
Psi = create_psi(BasisIndices, self._univ_p_val)
+ #print(Psi.shape)
+
# Calulate the cofficients of the metamodel
outs = self.regression(Psi, ED_Y, BasisIndices)
@@ -817,14 +812,12 @@ class PCE(MetaModel):
method='user'
)
# Compute univariate bases for the given samples
- _univ_p_val = None
_univ_p_val = self.univ_basis_vals(
samples,
n_max=np.max(self._pce_deg)
)
+ #print(_univ_p_val)
- mean_pred = None
- std_pred = None
mean_pred_b = {}
std_pred_b = {}
b_i = 0
@@ -847,6 +840,7 @@ class PCE(MetaModel):
# Assemble Psi matrix
basis = self._basis_dict[f'b_{b_i + 1}'][output][in_key]
psi = create_psi(basis, _univ_p_val)
+ #print(basis)
#print(psi.shape)
@@ -854,6 +848,10 @@ class PCE(MetaModel):
if self.bootstrap_method != 'fast' or b_i == 0:
# with error bar, i.e. use _clf_poly
_clf_poly = self._clf_poly[f'b_{b_i + 1}'][output][in_key]
+ #import inspect
+ #args = inspect.getfullargspec(_clf_poly.predict)
+ # TODO: check if 'return_std' in this
+ #if args.varargs is not None and 'return_std' in args.varargs:
try:
y_mean, y_std = _clf_poly.predict(
psi, return_std=True
@@ -861,6 +859,9 @@ class PCE(MetaModel):
except TypeError:
y_mean = _clf_poly.predict(psi)
y_std = np.zeros_like(y_mean)
+ #else:
+ # y_mean = _clf_poly.predict(psi)
+ # y_std = np.zeros_like(y_mean)
else:
# without error bar
coeffs = self._coeffs_dict[f'b_{b_i + 1}'][output][in_key]
diff --git a/src/bayesvalidrox/surrogate_models/sequential_design.py b/src/bayesvalidrox/surrogate_models/sequential_design.py
index a373f8dd16a716a670478ad2b1bf044619327290..38c5c3ac9df71c573bac6114aaa61e169e8718dc 100644
--- a/src/bayesvalidrox/surrogate_models/sequential_design.py
+++ b/src/bayesvalidrox/surrogate_models/sequential_design.py
@@ -1125,7 +1125,7 @@ class SequentialDesign:
# TODO: Only one psi can be selected.
# Suggestion: Go for the one with the highest LOO error
# TODO: this is just a patch, need to look at again!
- Scores = list(self.MetaModel.score_dict['b_1'][OutputName].values())
+ Scores = list(self.MetaModel.LOOCV_score_dict['b_1'][OutputName].values())
ModifiedLOO = [1 - score for score in Scores]
outIdx = np.argmax(ModifiedLOO)
@@ -1535,7 +1535,7 @@ class SequentialDesign:
"""
# Compute the mean and std based on the MetaModel
- pce_means, pce_stds = self.MetaModel._compute_pce_moments()
+ pce_means, pce_stds = self.MetaModel.compute_moments()
# Compute the root mean squared error
for output in self.out_names:
diff --git a/src/bayesvalidrox/surrogate_models/surrogate_models.py b/src/bayesvalidrox/surrogate_models/surrogate_models.py
index a8537f66692a80cbc26915839be271663e28468f..49733129f35c6c8d601b3ec00820094f1fbc3ac4 100644
--- a/src/bayesvalidrox/surrogate_models/surrogate_models.py
+++ b/src/bayesvalidrox/surrogate_models/surrogate_models.py
@@ -94,12 +94,11 @@ class MetaModel:
self.CollocationPoints = None # TODO: rename this
self.pca = None
self.LCerror = None
- #self.score_dict = None
- #self.basis_dict = None
+ self.LOOCV_score_dict = None # TODO: generic function that calculates this score!?
#self.x_scaler = None
#self.gp_poly = None
self.n_params = None
- self.ndim = None
+ self.ndim = None # TODO: this should be the same as n_params!?
#self.init_type = None # TODO: this parameter seems to not be used
#self.rmse = None # TODO: check if this is needed