A computational framework for mixture and composite distributions.
📦 Flexible Distribution Modeling
- Comprehensive support for standard, composite, and mixture distributions.
- Built-in symbolic differentiation engine for transformations and inverse transformations, simplifying complex operations on random variables.
📊 Advanced Statistical Functions
- Seamlessly evaluates PDFs, CDFs, quantiles, and random samples.
- Includes monotonic transformations and mixtures with custom weights.
🧮 Efficient Mathematical Framework
- Automatic construction of required equations for transformed random variables.
- Supports truncated and hierarchical distributions for specialized modeling.
📖 Extensive Documentation
- Detailed vignettes and examples, including tutorials on risk measures, parameter estimation, and real-world applications.
mistr is a cutting-edge R package that enables users to define, manipulate, and analyze univariate probability distributions. Whether modeling complex financial data or designing custom distributions, mistr provides the tools needed for a wide range of statistical tasks.
Highlights include:
- Composite and Mixture Models: Easily construct and analyze hierarchical and truncated distributions.
- Symbolic Differentiation: Built-in engine to derive and manage equations for transformations and inversions of random variables.
- Versatility: From academic research to real-world applications,
mistrsimplifies handling distributions.
- Financial Mathematics: Model heavy-tailed data for asset returns, risk management, and loss modeling.
- Actuarial Science: Fit and evaluate composite models for insurance claim distributions.
- Custom Distributions: Create and analyze tailored distributions for experimental data.
If you use mistr in your research, please cite:
@article{mistr2020,
title = "{mistr: A Computational Framework for Mixture and Composite Distributions}",
author = {Lukas Sablica and Kurt Hornik},
journal = "{The R Journal}",
year = "2020",
volume = "12",
number = "1",
pages = "283--299",
doi = "10.32614/RJ-2020-003",
url = "https://journal.r-project.org/archive/2020/RJ-2020-003/index.html"
}