README.md

In MiguelRodo/gamlssutils:

gamlssutils

The goal of gamlssutils is to provide functions to ease extracting and displaying output from GAMLSS models.

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("MiguelRodo/gamlssutils")

Example

Here we load the package and create example data:

library(gamlssutils)
set.seed(1)
data_mod <- data.frame(x = rnorm(20, 5) + rgamma(20, 0.01, 0.0001))
data_mod[['y']] <- 3 * data_mod[['x']] + rnorm(20, sd = 5)
head(data_mod)
#>          x        y
#> 1 4.373585 18.90344
#> 2 5.183644 19.71117
#> 3 4.164371 11.35647
#> 4 6.595281 21.11653
#> 5 5.329508 14.10501
#> 6 4.179532 24.74542

You can easily extract coefficients from all parameters into an organised table:

mod <- gamlss::gamlss(formula = y ~ -1 +  x,
                      family = "BCPE",
                      data = data_mod,
                      control = gamlss::gamlss.control(trace = FALSE))
get_coef_tbl(mod = mod)
#> # A tibble: 2 x 5
#>   Variable  `Mu coefficient` `Sigma coefficie~ `Nu coefficient` `Tau coefficien~
#>   <chr>                <dbl>             <dbl>            <dbl>            <dbl>
#> 1 Intercept            NA                0.389            0.796            0.841
#> 2 x                     2.87            NA               NA               NA

These tables are pleasantly displayed with pander::pandoc.table (remember to set results = 'asis' in chunk options):

# install.packages('pander')
pander::pandoc.table(get_coef_tbl(mod = mod))

| Variable | Mu coefficient | Sigma coefficient | Nu coefficient | Tau coefficient | | :-------: | :------------: | :---------------: | :------------: | :-------------: | | Intercept | NA | 0.3889 | 0.7964 | 0.8413 | | x | 2.868 | NA | NA | NA |

You can get the LR test results neatly put into a table:

# data
set.seed(1)
data_mod <- data.frame(x = rnorm(20, 5) + rgamma(20, 0.01, 0.0001))
data_mod[['y']] <- 3 * data_mod[['x']] + rnorm(20, sd = 5)

# models
mod_mu <- gamlss::gamlss(formula = y ~ -1 +  x,
                          family = "NO",
                          data = data_mod,
                          control = gamlss::gamlss.control(trace = FALSE))
mod_sigma <- gamlss::gamlss(formula = y ~ -1 +  x,
                            sigma.formula =  ~ x,
                            family = "NO",
                            data = data_mod,
                            control = gamlss::gamlss.control(trace = FALSE))
# table
lr_tbl <- get_lr_tbl(mod = mod_sigma, mod_null = mod_mu, 
                     mod_null_name = "sigma")
pander::pandoc.table(lr_tbl)

| Comp. model | LR Chi^2 | LR DF | LR P-value | | :---------: | :------: | :---: | :--------: | | sigma | 3.916 | 1 | 0.04784 |

The function (i.e. the ) method for objects) expects to find the object passed to the parameter in the call to to be found in the global environment. However, this may often not be the case, as that object may have changed or else may longer/may have never existed in the global environment. The function get_vcov gets around it. You pass it the dataframe that was given to the function and the name it had at the time (as a string), and get_vcov will then be able to calculate the variance-covariance matrix in any environment.

MiguelRodo/gamlssutils documentation built on July 9, 2020, 12:48 a.m.