| Title: | Robust Latent Profile Analysis |
| Version: | 0.1.0 |
| Description: | Provides a comprehensive toolset for estimating Latent Profile Analysis (LPA) models that are robust to multivariate outliers and missing data. By integrating a high-performance 'C++' engine via 'RcppArmadillo' with a Full Information Maximum Likelihood (FIML) approach and Huber weighting, it reliably extracts latent profiles even in complex datasets. It supports multiple geometric variance-covariance models, along with functions for bootstrapped likelihood ratio tests and plotting. For methodological details on the Bootstrapped Likelihood Ratio Test, see Nylund et al. (2007) <doi:10.1080/10705510701575396>. For robust clustering methods, see Garcia-Escudero et al. (2010) <doi:10.1007/s11634-010-0064-5>. |
| License: | GPL (≥ 3) |
| Encoding: | UTF-8 |
| RoxygenNote: | 8.0.0 |
| LinkingTo: | Rcpp, RcppArmadillo |
| Imports: | Rcpp, ggplot2, stats |
| NeedsCompilation: | yes |
| Packaged: | 2026-06-30 08:24:49 UTC; hp |
| Author: | Valerio Riccardo Aquila
 [aut, cre] |
| Maintainer: | Valerio Riccardo Aquila <valerio_aquila@hotmail.it> |
| Repository: | CRAN |
| Date/Publication: | 2026-07-05 11:00:02 UTC |
RobustLPA: Robust Latent Profile Analysis
Description
Provides a comprehensive toolset for estimating Latent Profile Analysis (LPA) models that are robust to multivariate outliers and missing data. By integrating a high-performance 'C++' engine via 'RcppArmadillo' with a Full Information Maximum Likelihood (FIML) approach and Huber weighting, it reliably extracts latent profiles even in complex datasets. It supports multiple geometric variance-covariance models, along with functions for bootstrapped likelihood ratio tests and plotting. For methodological details on the Bootstrapped Likelihood Ratio Test, see Nylund et al. (2007) doi:10.1080/10705510701575396. For robust clustering methods, see Garcia-Escudero et al. (2010) doi:10.1007/s11634-010-0064-5.
Author(s)
Maintainer: Valerio Riccardo Aquila valerio_aquila@hotmail.it (ORCID)
Authors:
Valerio Riccardo Aquila valerio_aquila@hotmail.it (ORCID)
Bootstrapped Likelihood Ratio Test for Robust LPA
Description
Compares a robust LPA model with G profiles against a null model with G-1 profiles using parametric bootstrapping. Supports FIML simulation conditions.
Usage
blrt_robust(data, G, model = 6, n_samples = 50, n_starts = 2)
Arguments
data |
A matrix or data.frame. |
G |
The number of profiles for the alternative hypothesis (compared against G-1). |
model |
An integer (1 to 6) specifying the variance-covariance parameterization. |
n_samples |
Number of bootstrap samples. Default is 50 for speed; 200+ is recommended for publications. |
n_starts |
Number of starts for the EM algorithm execution. |
Value
A list containing the observed LRT statistic, the vector of bootstrap replicates, and the empirical p-value.
Examples
# Fast demonstration of the robust BLRT
data(iris)
blrt_res <- blrt_robust(iris[1:30, 1:2], G = 2, model = 1, n_samples = 2, n_starts = 1)
# Print the summary of the results
blrt_res
Estimate Robust Latent Profile Models
Description
This function provides a user-friendly interface to fit multiple robust LPA models simultaneously across different numbers of profiles and model structures, returning a comprehensive fit summary table.
Usage
estimate_profiles_robust(
data,
n_profiles = 1:3,
models = c(1, 2, 3, 4, 5, 6),
n_starts = 5
)
Arguments
data |
A matrix or data.frame. |
n_profiles |
A vector of integers specifying the number of profiles to run (e.g., 1:3). |
models |
A vector of LPA models to run (e.g., c(1, 2, 3, 4, 5, 6)). Default is c(1, 2, 3, 4, 5, 6). |
n_starts |
Number of initializations per model. |
Value
A list containing the fit comparison table and the estimated models.
Examples
# Quick evaluation of multiple profiles
data(iris)
res <- estimate_profiles_robust(iris[1:30, 1:2], n_profiles = 1:2, models = 1, n_starts = 1)
res$fit_table
Plot Robust Latent Profiles
Description
Automatically generates a professional profile plot using ggplot2 from an estimated robust LPA model.
Usage
plot_robust_lpa(
model,
title = "Robust Latent Profiles",
xlab = "Variables",
ylab = "Value",
var_labels = NULL,
legend_title = "Class"
)
Arguments
model |
A model object returned by robust_lpa or estimate_profiles_robust. |
title |
The title of the plot. Default is "Robust Latent Profiles". |
xlab |
The x-axis label. Default is "Variables". |
ylab |
The y-axis label. Default is "Value". |
var_labels |
A character vector to manually rename the variables on the X axis. Default is NULL (auto-detect). |
legend_title |
The title of the legend. Default is "Class". |
Value
A ggplot object.
Examples
data(iris)
fit <- robust_lpa(data = iris[1:30, 1:2], G = 2, model = 1, n_starts = 1)
plot_robust_lpa(fit)
Fit a Single Robust Latent Profile Analysis Model
Description
This function estimates a single robust Latent Profile Analysis (LPA) model for a specified number of profiles and model structure. It automatically handles missing data via robust FIML if present.
Usage
robust_lpa(data, G, model = 6, max_iter = 100, tol = 1e-06, n_starts = 5)
Arguments
data |
A matrix or data.frame of observations. |
G |
The number of latent profiles to extract. |
model |
An integer (1 to 6) specifying the model parameterization. |
max_iter |
Maximum number of EM iterations. |
tol |
Tolerance for convergence. |
n_starts |
Number of random initializations. |
Value
A list containing parameters, fit indices, and assignments.
Examples
# Quick example using a small subset of the iris dataset
data(iris)
fit <- robust_lpa(data = iris[1:30, 1:2], G = 2, model = 1, n_starts = 1)
# Print the fit indices
fit$fit
Perform a Robust M-Step for Latent Profile Analysis
Description
Perform a Robust M-Step for Latent Profile Analysis
Usage
robust_m_step(data, z, alpha = 0.05)
Arguments
data |
A matrix or data.frame of observations. |
z |
A numeric vector containing the posterior probabilities of belonging to this cluster. |
alpha |
Significance level for the Chi-squared outlier cutoff (default is 0.05). |
Value
A list containing the robust 'mean' vector and the robust 'covariance' matrix.
Examples
data(iris)
# Simulate initial probabilities for a single profile
z_init <- runif(30)
z_init <- z_init / sum(z_init)
m_step_res <- robust_m_step(iris[1:30, 1:2], z = z_init)
# Print the robust mean calculated in the M-step
m_step_res$mean
Calculate the robust mean
Description
Calculate the robust mean
Usage
robust_mean(data, threshold = 10)
Arguments
data |
A matrix or data.frame |
threshold |
Maximum distance allowed to not be considered an outlier |
Value
A numeric vector representing the robust mean of the variables.
Examples
data(iris)
r_mean <- robust_mean(iris[1:30, 1:2])
# Print the calculated robust means
r_mean