README.md

In rdmatheus/bergrm: BerG Regression Model for Count Data

bergrm: BerG Regression Model for Count Data

Rodrigo M. R. Medeiros [e-mail], IME-USP

The bergrm package (in development) provide a set of functions for a complete regression analysis of count data, in which it is assumed that the dependent variable follows a BerG distribution. The BerG regression was proposed by Bourguignon and Medeiros (2020), and can be used to fit count data with overdispersion, equidispersion (but are not Poisson-distributed), and underdispersion.

Installation

You can install the current development version of bergrm from GitHub with:

devtools::install_github("rdmatheus/bergrm")

To run the above command, it is necessary that the devtools package is previously installed on R. If not, install it using the following command:

install.packages("devtools")

After installing the devtools package, if you are using Windows, install the most current RTools program. Finally, run the command devtools::install_github("rdmatheus/bergrm"), and then the package will be installed on your computer.

Example

This package provide complete estimation and inference for the parameters as well as simulation envelope plots, useful for assessing the goodness-of-fit of the model. The implementation is straightforward and similar to other popular packages, like betareg and glm, where the main function is glm.bg() (the main function was recently modified from bergrm to glm.bg). Below is an example of some functions usage and available methods.

library(bergrm)

There are currently two data sets available in the bergrm package, grazing and bids. To access the documentation associated with this data, use help(grazing) and help(bids), respectively. In this example we use the grazing data set.

# Data visualization (For description: ?grazing)

# Fit a double model (mean and dispersion) with a test for constant dispersion

fit <- glm.bg(birds ~ when + grazed | when + grazed, data = grazing, disp.test = TRUE)

## For glm.bg documentation:
?glm.bg

# Print

fit
#> 
#> Call:
#> glm.bg(formula = birds ~ when + grazed | when + grazed, data = grazing, 
#>     disp_test = TRUE)
#> 
#> mean Coefficients:
#> [1]  2.2093390  0.2780493 -0.7752226
#> 
#> dispersion Coefficients:
#> [1]  2.42884684 -0.02824461 -0.43131213

# Summary

summary(fit)
#>
#> Call:
#> glm.bg(formula = birds ~ when + grazed | when + grazed, data = grazing, 
#>        disp_test = TRUE)
#>
#> Summary for residuals:
#>        Mean       Sd  Skewness Kurtosis
#>    0.013563 0.982888  0.080752 3.057912
#> 
#> ----------------------------------------------------------------
#> Mean:
#> Coefficients:
#>             Estimate Std. Error z value  Pr(>|z|)    
#> (Intercept)  2.20934    0.31300  7.0586 1.682e-12 ***
#> whenAfter    0.27805    0.32385  0.8586   0.39058    
#> grazedFeral -0.77522    0.32582 -2.3793   0.01734 *  
#> ---
#> Signif. codes:  
#>   0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> 
#> ----------------------------------------------------------------
#> Dispersion:
#> 
#> Link function: log 
#> Coefficients:
#>              Estimate Std. Error z value  Pr(>|z|)    
#> (Intercept)  2.428847   0.316975  7.6626 1.822e-14 ***
#> whenAfter   -0.028245   0.331138 -0.0853    0.9320    
#> grazedFeral -0.431312   0.339611 -1.2700    0.2041    
#> ---
#> Signif. codes:  
#>   0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’  0.1 ‘ ’ 1
#> 
#> ----------------------------------------------------------------
#> 
#> Test for constant dispersion:
#>              S      W     LR      G
#> Value   2.5045 1.6195 1.9735 1.9987
#> P value 0.2859 0.4450 0.3728 0.3681
#> 
#> In addition, Log-lik value: -178.36 
#> AIC: 368.72 and BIC: 381.4828

# Plot

plot(fit, residual = "quantile") 

## Use residual = "pearson" to obtain a summary for the Pearson residuals.

# Normal probability plot with simulated envelope 

envel_berg(fit)

## Here we can also choose the residual using residual = "quantile" for
## the randomized quantile residuals; "pearson" for Pearson residuals;
## and "all" (default) for both.

# Rootogram

root_berg(fit)

Currently, the methods implemented for "bergrm" objects are

methods(class = "bergrm")
#>
#> [1] AIC          coef         logLik       model.matrix
#> [5] plot         print        summary      vcov        
#> see '?methods' for accessing help and source code

rdmatheus/bergrm documentation built on Oct. 1, 2020, 4:38 a.m.