modeltools

The goal of modeltools is to provide small, reusable tools for econometric modeling in R, with a particular focus on choice models estimated with bgw. Model objects work with the stats, broom and sandwich generics, so standard tools such as AIC(), tidy() and sandwich() work out of the box.

Installation

You can install the development version of modeltools from GitHub with:

# install.packages("devtools")
devtools::install_github("edsandorf/modeltools")

Example

Robust standard errors for a model estimated with bgw_mle():

library(modeltools)

model <- bgw::bgw_mle(
  log_lik, 
  betaStart = beta, 
  bgw_settings = list(vcHessianMethod = "finiteDifferences")
)
model <- add_scores(model, log_lik)

tidy(model, vcov = sandwich(model))
prep_for_gt(model, vcov = vcovCL(model, cluster = db$id))

Comparing two empirical distributions using the Poe et al. (2005) test:

library(modeltools)

set.seed(123)
wtp_a <- rnorm(1000, mean = 0.5)
wtp_b <- rnorm(1000, mean = 1)
poe_test(wtp_a, wtp_b)
#> Method:  Poe et al. (2005) test 
#> 
#> Means: 
#> wtp_a wtp_b 
#> 0.516 1.042 
#> 
#> H0: x = y 
#> H1: x > y | x < y 
#> 
#> Gamma:  0.646704 
#> 
#> Gamma >.95 and <.05 indicates difference at the 5% level.