README.md

In marcosop/HeckmanEM: Fit Normal, Student-t or Contaminated Normal Heckman Selection Models

HeckmanEM: Fit Normal, Student-t or Contaminated normal Heckman Selection Models

Overview

It performs maximum likelihood estimation for the Heckman selection model (Normal, Student-t or Contaminated normal) using an EM-algorithm (Lachos et al., 2021). It also performs influence diagnostic through global and local influence for four possible perturbation schema.

Installation

You can install HeckmanEM from CRAN when available, or you can install the development version from GitHub with the following commands:

# Install from CRAN (when available)
install.packages("HeckmanEM")

# Or the development version from GitHub
# install.packages("devtools")
devtools::install_github("marcosop/HeckmanEM")

Available Functions

The HeckmanEM package provides tools for fitting and analyzing the Heckman Selection model. Here is a list of the functions available in the package:

• HeckmanEM(): Fits a Heckman Selection model, adjusting for potential selection bias.

• HeckmanEM.criteria(): Calculates the AIC, AICc, and BIC model selection criteria for a fitted Heckman Selection model.

• HeckmanEM.envelope(): Generates an envelope plot for a fitted Heckman Selection model, providing a visual model fit assessment.

• HeckmanEM.infomat(): Estimates standard errors for the parameters of a fitted model, quantifying parameter uncertainty.

• Influence(): Performs an influence analysis for a HeckmanEM object to assess the effect of excluding individual observations.

• CaseDeletion(): Carries out case deletion analysis for a HeckmanEM object, studying the impact of observation removal on parameter estimates.

• rHeckman(): Generates a random sample from a Heckman Selection model (Normal or Student-t), facilitating data simulation.

Each of these functions comes with complete documentation, including examples of usage. You can access this documentation using the help() function in R.

Tutorial

# First, let's generate some data from the Heckman Selection model.

n    <- 100
nu   <- 3
cens <- 0.25

# We set a seed to make the results reproducible
set.seed(13)

# We generate the covariates matrices
w <- cbind(1, runif(n, -1, 1), rnorm(n))
x <- cbind(w[,1:2])

# We calculate the quantile of the Student's T distribution
c <- qt(cens, df = nu)

# We set the values of the parameters
sigma2   <- 1
beta     <- c(1, 0.5)
gamma    <- c(1, 0.3, -.5)
gamma[1] <- -c * sqrt(sigma2)

# We generate the data
datas <- rHeckman(x, w, beta, gamma, sigma2, rho = 0.6, nu, family = "T")
y     <- datas$y
cc    <- datas$cc

# Now, let's fit a Heckman Selection model to the data.
heckmodel <- HeckmanEM(y, x, w, cc, family = "Normal", iter.max = 50)

# We perform a case deletion analysis on the fitted model.
global <- CaseDeletion(heckmodel)

# We create a plot of the result.
plot(global)

# We perform an influence analysis with different types of perturbations.

# Case when the weight of an observation is altered
local_case <- Influence(heckmodel, type = "case-weight")
# Influential observations
local_case$influent 
# Visualization
plot(local_case)

# Case when an observation is scaled
local_scale <- Influence(heckmodel, type = "scale")
# Influential observations
local_scale$influent
# Visualization
plot(local_scale)

# Case when the response of an observation is altered
local_response <- Influence(heckmodel, type = "response")
# Influential observations
local_response$influent
# Visualization
plot(local_response)

# Exploratory case in which a covariate is altered (column 2 in this case)
local_explore <- Influence(heckmodel, type = "exploratory", colx = 2)
# Influential observations
local_explore$influent
# Visualization
plot(local_explore)

Authors

This package has been developed by:

• Marcos Prates (Author, Creator, Translator), email: [e-mail]
• Victor Lachos (Author)
• Dipak Dey (Author)
• Marcos Oliveira (Author, Contributor)
• Christian Galarza (Contributor)
• Katherine Loor (Contributor)
• Alejandro Ordoñez (Contributor)

Contributions

Contributions to the software are welcome. If you find a bug or have an idea for an enhancement, please open an issue or submit a pull request.

Contact

If you have any questions, feel free to open an issue or send an email to the repository owner.

License

This software is open source under the MIT License. Please see the LICENSE file for more information.

marcosop/HeckmanEM documentation built on July 17, 2025, 3:20 a.m.