mpest.initializers#
initializers module for mixture model parameter initialization.
This module provides various strategies for initializing parameters of mixture models before the main optimization process. Good initialization is crucial for achieving fast convergence and high-quality solutions in mixture model estimation.
Purpose
initializers provide good starting points for EM algorithm and other optimization methods, helping to avoid poor local optima and improving convergence.
Usage Example
>>> from rework_pysatl_mpest import Exponential
>>> import numpy as np
>>> from sklearn.cluster import KMeans
>>> from rework_pysatl_mpest.initializers import ClusterizeInitializer
>>> from rework_pysatl_mpest.initializers import ClusterMatchStrategy, EstimationStrategy
>>> # Create initializer with KMeans clustering
>>> initializer_cluster = ClusterizeInitializer(
... is_accurate=True,
... is_soft=False,
... clusterizer=KMeans(n_clusters=3)
... )
>>> # Create distribution models to initialize
>>> distributions = [Exponential(loc=0.0, rate=0.1),
>>>Exponential(loc=5.0, rate=0.05), Exponential(loc=10.0, rate=0.01)]
>>> # Generate sample data
>>> X = np.linspace(0.01, 25.0, 300)
>>> # Perform initialization
>>> mixture_model = initializer_cluster.perform(
... X=X,
... dists=distributions,
... cluster_match_strategy=ClusterMatchStrategy.AKAIKE,
... estimation_strategies=[EstimationStrategy.QFUNCTION] * len(distributions)
... )
>>> # The mixture model is now initialized with estimated parameters
>>> print(f"Number of components: {len(mixture_model.components)}")
>>> print(f"Weights: {mixture_model.weights}")
Initialization Strategies#
Abstract Classes#
Abstract base class for mixture model initializers. |
Concrete Implementations#
Cluster-based initializer for mixture model parameters. |
Strategy Enumerations#
Enumeration types that define available strategies for initialization.
Enumeration of strategies for matching clusters to distribution models. |
|
Enumeration of parameter estimation strategies for distribution components. |