The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

EstemPMM News

Version 0.3.1 (2026-04-06)

CRAN Resubmission


Version 0.3.0 (2026-03-19)

New Feature: PMM3 for Symmetric Platykurtic Errors

PMM3 (S=3) extends the Polynomial Maximization Method to handle symmetric error distributions with negative excess kurtosis (platykurtic), such as uniform, beta-symmetric, and truncated normal errors.

New Functions

PMM3 Time Series Functions

New S4 Classes

Documentation


Version 0.2.0 (2025-11-20)

Major Update: Unified PMM2 Architecture

This release represents a significant architectural improvement based on comprehensive research comparing different PMM2 implementation strategies.

New Features

Research-Based Improvements

Based on Monte Carlo simulations (R=50, n=200) comparing three approaches:

Approach AR(1) MA(1) SARIMA Status
Unified Iterative -2.9% MSE -19.9% MSE -16.4% MSE Best overall
Unified One-step -2.2% MSE -23.0% MSE -15.6% MSE Fastest
Linearized (MA) N/A -21.6% MSE N/A MA specialist
Direct Nonlinear N/A Failed Failed Removed

Key findings: - Unified approaches provide consistent 3-23% MSE improvement - One-step (global) variant offers best speed/accuracy tradeoff - Linearized approach optimal for pure MA/SMA models

Breaking Changes

API Changes

# Old way (still works, uses unified_global by default)
ar_pmm2(y, order = 2)

# New explicit variant selection
ar_pmm2(y, order = 2, pmm2_variant = "unified_iterative")
ma_pmm2(y, order = 1, pmm2_variant = "linearized")  # Best for MA
arima_pmm2(y, order = c(1,0,1), pmm2_variant = "unified_global")  # Default

Documentation

Dependencies

Bug Fixes

Performance


Version 0.1.4 (Development - Superseded by 0.2.0)

New Features

Bug Fixes

Version 0.1.3 (2025-11-13)

Documentation

Version 0.1.2 (2025-11-13)

New Features

Bug Fixes

Improvements

Version 0.1.1 (2025-10-23)

Maintenance

Version 0.1.0 (2025-01-15)

Initial Release: PMM2 Foundation

New Features: - lm_pmm2() - Linear regression estimation using Polynomial Maximization Method (S=2) - ar_pmm2() - Autoregressive (AR) time series modeling with PMM2 - ma_pmm2() - Moving Average (MA) time series modeling with PMM2 - arma_pmm2() - ARMA time series modeling with PMM2 - arima_pmm2() - ARIMA time series modeling with PMM2 - pmm2_inference() - Bootstrap inference for linear models - ts_pmm2_inference() - Bootstrap inference for time series models - Statistical utilities: pmm_skewness(), pmm_kurtosis(), compute_moments() - Comparison functions: compare_with_ols(), compare_ts_methods(), compare_ar_methods(), compare_ma_methods(), compare_arma_methods(), compare_arima_methods()

S4 Classes: - PMM2fit - Results container for linear regression models - TS2fit - Base class for time series results - ARPMM2, MAPMM2, ARMAPMM2, ARIMAPMM2 - Specialized time series result classes

Methods: - summary() - Model summary statistics - coef() - Extract coefficients - fitted() - Fitted values - predict() - Predictions for new data - residuals() - Model residuals - plot() - Diagnostic plots

Documentation: - Comprehensive Roxygen2 documentation for all exported functions - README with theoretical background and basic usage examples - Demonstration script pmm2_demo_runner.R showing practical applications

Package Architecture

Module Organization: - R/pmm2_main.R - Primary PMM2 fitting functions - R/pmm2_classes.R - S4 class definitions - R/pmm2_utils.R - Utility functions for moment computation and optimization - R/pmm2_ts_design.R - Time series design matrix construction

Dependencies: - Core: methods, stats, graphics, utils - Optional: MASS (for advanced statistical functions, available in Suggests)

Quality Assurance: - Unit tests covering core PMM2 functionality - Edge case handling for numerical stability - Convergence diagnostics and warnings

Known Limitations

Roadmap

1.0.0 (Stable API): - API stabilization and backward compatibility guarantee - Seasonal PMM3 models (sar_pmm3, sarima_pmm3) - Extended performance benchmarks - Specialized applications (econometrics, biostatistics)

Citation

If you use EstemPMM in your research, please cite the relevant publications:

For Linear Regression (lm_pmm2): Zabolotnii S., Warsza Z.L., Tkachenko O. (2018) Polynomial Estimation of Linear Regression Parameters for the Asymmetric PDF of Errors. In: Szewczyk R., Zieliński C., Kaliczyńska M. (eds) Automation 2018. AUTOMATION 2018. Advances in Intelligent Systems and Computing, vol 743. Springer, Cham. https://doi.org/10.1007/978-3-319-77179-3_75

For Autoregressive Models (ar_pmm2): Zabolotnii S., Tkachenko O., Warsza Z.L. (2022) Application of the Polynomial Maximization Method for Estimation Parameters of Autoregressive Models with Asymmetric Innovations. In: Szewczyk R., Zieliński C., Kaliczyńska M. (eds) Automation 2022. AUTOMATION 2022. Advances in Intelligent Systems and Computing, vol 1427. Springer, Cham. https://doi.org/10.1007/978-3-031-03502-9_37

For Moving Average Models (ma_pmm2): Zabolotnii S., Tkachenko O., Warsza Z.L. (2023) Polynomial Maximization Method for Estimation Parameters of Asymmetric Non-gaussian Moving Average Models. In: Szewczyk R., et al. (eds) Automation 2023. AUTOMATION 2023. Lecture Notes in Networks and Systems, vol 630. Springer, Cham.

Technical Notes

Algorithm Stability: - Regularization parameter automatically adjusted for ill-conditioned systems - Step size limiting prevents divergence in optimization - Convergence history tracking for diagnostics

Numerical Considerations: - Moment estimation uses robust methods to handle outliers - Design matrices constructed with numerical stability in mind - NA/Inf values detected and handled appropriately

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.