ecdf_estimator.utils Namespace Reference

Functions
def	empirical_cumulative_distribution_vector (distance_list, bins)
	Create list of ECDF values. More...
def	create_distance_matrix (dataset_list, distance_fct, start_index_list=None, end_index_list=None)
	Assemble list of (generalized) distances between elemenst of datasets. More...
def	empirical_cumulative_distribution_vector_list (dataset, bins, distance_fct, subset_indices, compare_all=True)
	Assemble ecdf vector, whose elements are list of values for all subset combinations. More...
def	empirical_cumulative_distribution_vector_list_bootstrap (dataset, bins, distance_fct, n_elements_a, n_elements_b, n_samples)
	Same as empirical_cumulative_distribution_vector_list, but for bootstrapping. More...
def	mean_of_ecdf_vectors (ecdf_vector_list)
	Vector of means of the ecdf vectors. More...
def	covariance_of_ecdf_vectors (ecdf_vector_list)
	Covariance matrix of ecdf vectors. More...
def	evaluate (estimator, dataset)
	Evaluate target/objective/cost function associated to estimator type from dataset. More...
def	evaluate_from_empirical_cumulative_distribution_functions (estimator, vector)
	Evaluate target/objective/cost function associated to estimator type from ecdf vector. More...

Function Documentation

covariance_of_ecdf_vectors()

def ecdf_estimator.utils.covariance_of_ecdf_vectors

(

ecdf_vector_list

)

Covariance matrix of ecdf vectors.

Parameters

ecdf_list

Usually the result of empirical_cumulative_distribution_vector_list.

Return values

ecdf_covar

Covariance matrix associated to the ecdf vectors.

create_distance_matrix()

def ecdf_estimator.utils.create_distance_matrix	(		dataset_list,
			distance_fct,
			start_index_list = None,
			end_index_list = None
	)

Assemble list of (generalized) distances between elemenst of datasets.

This function creates a list of distances between elements of subsets which are compared to each other. The amount of subsets compared to each other is defined by the number of arguments in the distance function. The distances of all value combinations of these subsets are computed.

Parameters

dataset_list	The list of datasets which are compared to each other.
distance_fct	Generalized distance function among several datasets.
start_index_list	Starting indices of the datasets. Defaults to None.
end_index_list	Ending indices of the datasets. Defaults to None.

Return values

distance_mat

Matrix of generalized distances.

empirical_cumulative_distribution_vector()

def ecdf_estimator.utils.empirical_cumulative_distribution_vector	(		distance_list,
			bins
	)

Create list of ECDF values.

This function creates the empirical cumulative distribution functions from a list of distances and a list of bin values. That is, each element of the resulting list tells how many elements of the distance list are smaller than the respective bin value.

Parameters

distance_list	List of distances to be grouped into the bins.
bins	List of bins.

Return values

ecdf_list

Resulting list of amout of distances that are smaller than resp. bins.

empirical_cumulative_distribution_vector_list()

def ecdf_estimator.utils.empirical_cumulative_distribution_vector_list	(		dataset,
			bins,
			distance_fct,
			subset_indices,
			compare_all = True
	)

Assemble ecdf vector, whose elements are list of values for all subset combinations.

This function assembles a list of ecdf vectors for all possible combinations of subsets of the dataset. Importantly, none of the subsets are compared to themselves. The subset combinations and their indexes are chosen. Then, the distances are computed by create_distance_matrix function and added to a matrix.

The first dimension of the result refers to the index of the bin / ecdf vector. The second index of the result refers to the subset combination.

Parameters

dataset	Whole dataset, whose subset are compared to one another.
bins	List of bins.
distance_fct	Function generating a generalized distance between members of dataset.
subset_indices	List of starting (and ending) indices of disjointly subdivided dataset.
compare_all	If False, only subsets of different sizes are compared. Deafault: True

Return values

ecdf_list

ecdf vector enlisting values for subset combinations.

Here is the call graph for this function:

empirical_cumulative_distribution_vector_list_bootstrap()

def ecdf_estimator.utils.empirical_cumulative_distribution_vector_list_bootstrap	(		dataset,
			bins,
			distance_fct,
			n_elements_a,
			n_elements_b,
			n_samples
	)

Same as empirical_cumulative_distribution_vector_list, but for bootstrapping.

Parameters

dataset	Dataset, whose elements are compared to one another.
n_elements_a	Number of elements in first (smaller) subset.
n_elements_b	Number of elements in second (larger) subset.
bins	List of bins.
distance_fct	Function generating a generalized distance between members of dataset.
n_samples	Number of perturbatins of the datasets.

Return values

ecdf_list

ecdf vector enlisting values for subset combinations.

Here is the call graph for this function:

evaluate()

def ecdf_estimator.utils.evaluate	(		estimator,
			dataset
	)

Evaluate target/objective/cost function associated to estimator type from dataset.

Evaluate the negative log-likelihood in the way that is characterized by the estimator.

Parameters

estimator	The estimator class defining the specifics of the target function.
dataset	The dataset with respect to which the target function is evaluated.

Return values

target_val

The value of the target function.

evaluate_from_empirical_cumulative_distribution_functions()

def ecdf_estimator.utils.evaluate_from_empirical_cumulative_distribution_functions	(		estimator,
			vector
	)

Evaluate target/objective/cost function associated to estimator type from ecdf vector.

Evaluate the negative log-likelihood in the way that is characterized by the estimator.

Parameters

estimator	The estimator class defining the specifics of the target function.
ecdf_vector	The vector of ecdf, which is the argument for the target function.

Return values

target_val

The value of the target function.

mean_of_ecdf_vectors()

def ecdf_estimator.utils.mean_of_ecdf_vectors

(

ecdf_vector_list

)

Vector of means of the ecdf vectors.

Parameters

ecdf_list

Usually the result of empirical_cumulative_distribution_vector_list.

Return values

ecdf_means

Element-wise mean values of the ecdf vectors.