pcsclr: Progressive Censoring Schemes with Competitive Latent-Risk

The pcsclr package provides high-performance computing utilities for parameter estimation, simulation, and survival analysis under Progressive Censoring Schemes (PCS) in the presence of Competitive Latent Risks (CLR).

The package is designed to implement the methodologies developed in accompanying research manuscripts, providing both fast numerical maximum likelihood estimation (MLE) and robust Bayesian Markov Chain Monte Carlo (MCMC) samplers.

Features

• Numerical Optimization: Implements a 4th-order Runge-Kutta (RK4) path optimization framework to maximize log-likelihood structures and evaluate the observed Fisher Information matrix.
• High-Speed Bayesian Samplers: Features an accelerated C++ MCMC engine built via RcppArmadillo to execute random-walk Metropolis-Hastings update loops for Weibull baseline parameters under standard and hybrid progressive censoring setups.
• Data Simulation: Provides robust structures to simulate complex lifetime trajectories matching custom progressive removal patterns and multi-cause latent risk dropouts.

Installation

You can install the development version of pcsclr directly from GitHub using the remotes package. Run the following commands in your R console:

```R # Install remotes if you haven’t already if (!requireNamespace(“remotes”, quietly = TRUE)) { install.packages(“remotes”) }

Install pcsclr from the cloud repository

remotes::install_github(“https://github.com/obulezi12345-svg/pcsclr”)

Computational Architecture

The underlying architecture relies on a hybrid design to optimize memory allocation and computational speeds: * R Interface: Handles user inputs, input validations, configuration setups, and method dispatching. * C++ Core Engine: Evaluates joint-likelihood structures and structural boundary constraints (\([S(T_{max})]^{R^*}\)) inside highly tightly bound update loops to completely eliminate R expression evaluator execution overhead during heavy sampling profiles.

Contact & Citation

Author: Dr. Okechukwu J. Obulezi
Department: Department of Statistics, Nnamdi Azikiwe University, Awka, Nigeria
Email: oj.obulezi@unizik.edu.ng

Note: If you utilize this framework or package code profiles within academic publications, please cite the corresponding research paper referencing this implementation.