mpest.estimators#

Provides tools for estimating the parameters of mixture models.

This package brings together various algorithms and tools for estimating the parameters of MixtureModel objects from a data sample. All estimation algorithms inherit from the abstract base class BaseEstimator, which defines a unified interface—the fit method.

The package offers two main approaches to estimation:

  1. Iterative Algorithms (in the rework_pysatl_mpest.estimators.iterative submodule): A flexible module for building complex, multi-step algorithms like Expectation-Maximization (EM). It allows for fine-tuning each step, stopping conditions, and additional strategies.

  2. Direct Algorithms (in the rework_pysatl_mpest.estimators.direct submodule): Simpler methods that compute parameters in a single step, for example, by directly solving Maximum Likelihood Estimation (MLE) equations or using the Method of Moments (MoM).

For ease of use, Facades are also provided. These are high-level classes that hide the complexity of configuration and offer ready-to-use, popular algorithms.

Facades (High-Level API)#

ECM

An estimator that implements the Expectation-Conditional Maximization (ECM) algorithm.

Abstract Classes#

BaseEstimator

Abstract class for a mixture model parameter estimator.