| | |
- builtins.object
-
- SyntheticDataset
class SyntheticDataset(builtins.object) |
| |
SyntheticDataset(num_nodes: float, num_obs: float, node_labels: list, degree: float, graph_type: str = 'erdos-renyi', noise_scale: float = 1.0, true_dag=None)
Generate synthetic data from graphs.
|
| |
Methods defined here:
- __init__(self, num_nodes: float, num_obs: float, node_labels: list, degree: float, graph_type: str = 'erdos-renyi', noise_scale: float = 1.0, true_dag=None)
- Initialise SyntheticDataset instance
Static methods defined here:
- simulate_data(W: Union[numpy.ndarray, structure_learning.data_structures.dag.DAG], n, noise_scale=1.0, sigmas=None)
- Simulate samples from SEM with specified type of noise.
Parameters:
W (numpy.ndarray): weigthed DAG
n (int): number of samples
noise_scale (float): scale parameter of noise distribution in linear SEM
sigmas (numpy.ndarray): noise vector
Returns:
(numpy.ndarray) [n,d] sample matrix
- simulate_data_V1(W: Union[numpy.ndarray, structure_learning.data_structures.dag.DAG], n, noise_scale=1.0, sigmas=None)
- Simulate samples from SEM with specified type of noise.
Parameters:
W (numpy.ndarray): weigthed DAG
n (int): number of samples
noise_scale (float): scale parameter of noise distribution in linear SEM
sigmas (numpy.ndarray): noise vector
Returns:
(numpy.ndarray) [n,d] sample matrix
- simulate_data_from_dag(dag: Union[numpy.ndarray, structure_learning.data_structures.dag.DAG], num_obs, num_nodes, node_labels, w_range, noise_scale)
- Simulate samples from ground truth DAG.
Parameters:
DAG (numpy.ndarray): ground truth DAG
num_obs (int): number of observations to simulate
num_nodes (int): number of nodes
node_labels (list (str)): node labels
w_range (2-tuple (float)): weight range +/- (low, high)
noise_scale (float): scale parameter of noise distribution
Returns:
(numpy.ndarray): [n,d] sample matrix
- simulate_gaussian_dag(d, w_std)
- Simulate dense DAG adjacency matrix
Parameters:
d (int): number of nodes
w_std (float): weight std
Returns:
(numpy.ndarray): weighted DAG
None
(numpy.ndarray): permutation matrix
(numpy.ndarray): lower triangular matrix
- simulate_random_dag(d, degree, graph_type, w_range)
- Simulate random DAG with some expected degree.
Parameters:
d (int): number of nodes
degree (int): expected node degree, in + out
graph_type (str): {erdos-renyi, barabasi-albert, full}
w_range (2-tuple (float)): weight range +/- (low, high)
Returns:
(DAG): weighted DAG
None
(numpy.ndarray): permutation matrix
Data descriptors defined here:
- __dict__
- dictionary for instance variables
- __weakref__
- list of weak references to the object
| |