README.Rmd

---
output: github_document
---

<!-- README.md is generated from README.Rmd. Please edit that file -->

```{r setup, include=FALSE}

options(width = 100, tibble.print_max = 50)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
  )

```

<img src="man/figures/flexcm_hex.png" align="right" height=160/ display="block">

# `flexcm`: Flexible Models for Dispersed Count Data #

  [![Travis build status](https://travis-ci.org/jreduardo/flexcm.svg?branch=master)](https://travis-ci.org/jreduardo/flexcm)


> [Eduardo E. R. Junior](http://leg.ufpr.br/~eduardojr) -
> [jreduardo@usp.br](mailto:jreduardo@usp.br), IME-USP

The `flexcm` package contains functions to fit flexible count models
that can handle equi-, over-, and underdispersion, namely we consider
COM-Poisson, Gamma-count, discrete Weibull, generalized Poisson, double
Poisson and Poisson-Tweedie models^[Poisson-Tweedie models are fitted
using [`mcglm`][mcglm] package]. The normalizing constant for double
Poisson and COM-Poisson are written in C++.

Joint work with [Walmes M. Zeviani](www.leg.ufpr.br/~walmes/) and
[Clarice G.B. Demétrio](http://ce.esalq.usp.br/equipe/clarice-garcia-borges-demetrio).

## Installation

You can install the development version of `flexcm` from
[GitHub](https://github.com/jreduardo/flexcm) with:

```r

# install.packages("devtools")
devtools::install_github("jreduardo/flexcm")

```

## Usage and example

Basically, this package implements methods similar to those related to
glm objects. The main function is `flexcm(..., model)`.

```{r example}

library(flexcm)

# Fit models -----------------------------------------------------------

# Model families
families <- list("CMP" = "compoisson",
                 "GCT" = "gammacount",
                 "DWe" = "discreteweibull",
                 "GPo" = "generalizedpoisson",
                 "DPo" = "doublepoisson",
                 "PTw" = "poissontweedie")

# Fit models
models <- lapply(families, function(fam) {
  flexcm(ninsect ~ extract, model = fam, data = sitophilus)
})

# Methods --------------------------------------------------------------

vapply(models, coef, numeric(5))

vapply(models, logLik, numeric(1))

equitest(models[-3])

print(models[["PTw"]])

# Predict new data -----------------------------------------------------

newdf <- sitophilus[c(21, 31), -2, drop = FALSE]
purrr::map_dfr(models,
               .id = "model",
               .f = predict,
               newdata = newdf,
               type = "response",
               interval = "confidence",
               augment_data = TRUE)

```

Currently, the methods implemented for `"flexcm"` objects are

```{r methods}

methods(class = "flexcm")

```

## Related projects ##

There are other R packages to deal with COM-Poisson models that have
somehow contributed to the writing of `flexcm`.

- [`mcglm`][mcglm]: Routines for fitting the multivariate
  covariance genelralized linear models. Currently, this package is used
  to fit the Poisson-Tweedie models in `flexcm`.
- [`DWreg`][DWreg]: Fit Discrete weibull models (allows to model $q$ or
  $\rho$ parameters).
- [`gamlss.dist`][gamlss.dist]: Implements (among other) double Poisson
  and a restrictive generalized Poisson models.
- [`glmmTMB`][glmmTMB]: Fit (among other) COM-Poisson, under a different
  mean-parametrization, and generalized Poisson models (includes
  zero-inflation, dispersion modeling and random effects).

## License ##

The `flexcm` package is licensed under the [GNU General Public
License, version 3], see file `LICENSE.md`, © 2019 E. E., Ribeiro Jr.

<!------------------------------------------- -->
<!-- Links -->
[GNU General Public License, version 3]: https://www.gnu.org/licenses/gpl-3.0.ht
[mcglm]: https://github.com/wbonat/mcglm
[DWreg]: https://cran.r-project.org/package=DWreg
[glmmTMB]: https://github.com/glmmTMB/glmmTMB
[gamlss.dist]: https://cran.r-project.org/package=gamlss.dist