EstimationStrategy#
- class EstimationStrategy(value)[source]#
Bases:
EnumEnumeration of parameter estimation strategies for distribution components.
This enumeration defines the available methods for estimating parameters of individual distribution components during mixture model initialization.
- Variables:
QFUNCTION (EstimationStrategy) – Uses the Q-function (expected complete data log-likelihood) for parameter estimation. This strategy maximizes the Q-function either analytically (for specific distributions) or numerically (for general distributions).
Notes
QFUNCTION Strategy
The Q-function strategy: - For Exponential distribution: Uses analytical solution for efficiency - For other distributions: Falls back to numerical optimization - Provides maximum likelihood estimates in the EM framework - Handles weighted data points through the responsibility matrix
Future Extensions
Additional strategies that could be added: - Implementations for other types of distributions.