Graphical User Interface of ‘hbsaems’ Using ‘shiny’

The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

Introduction

The run_sae_app() function in the hbsaems package provides an interactive Shiny Dashboard for Hierarchical Bayesian Small Area Estimation (HBSAE) using brms for Bayesian inference with Stan. This application offers a user-friendly interface to upload data, define models, and obtain estimation results without requiring extensive R coding.

Install Required Packages

Ensure that you have installed the hbsaems package:

install.packages("hbsaems")

Load Required Packages

library(hbsaems)

Running the Shiny App

To launch the application, simply call:

run_sae_app()

This will start a Shiny application that runs in your web browser.

App Structure

1. Data Upload

Users can either upload a .csv file or select a data frame available in the current R environment.

Upload File: Choose and upload a .csv file from your computer.
Select from Environment: Choose an existing R data object.
Data Preview: The loaded data will be shown in a table preview for inspection.

2. Data Exploration

This tab provides four types of data exploration tools to help users understand the characteristics of the dataset:

Summary Statistics: Displays mean, median, min, max, and quartiles for selected numeric variables.
Histogram: Shows frequency distribution and density curve for selected variables.
Boxplot: Visualizes data spread, median, quartiles, and outliers.
Scatter Plot & Correlation: Visualizes relationships between two variables, with support for five correlation coefficients:
- Pearson
- Spearman’s Rho
- Chatterjee’s Xi
- Distance Correlation
- MIC (Maximal Information Coefficient)

3. Modeling

a. Modeling Configuration

Users can define key model components:

Basic Settings:
- Response Variable
- Auxiliary Variables (linear and nonlinear covariates)
- Group Variables (for hierarchical modeling)
- Distribution Type (e.g., Lognormal, Logitnormal, Beta, or Custom)
- HB Family & Link Function (for Custom models)
Spatial Modeling:
- Choose spatial type (SAR or CAR)
- Specify neighborhood structure
- Upload spatial weight matrix (.csv)
Missing Data Handling:
- Choose between deletion, imputation, or model-based handling

b. Prior Checking

Before fitting the model, users can configure prior distributions and perform prior predictive checks:

Summarize prior settings
Simulate from prior distributions
Visualize prior predictive plots

c. MCMC Settings

Configure sampling parameters for Bayesian estimation:

Seed
Chains
Cores
Thinning Rate
Iterations
Warm-up
Adapt Delta

Click “Fit Model” to begin model fitting using brms.

4. Results

After fitting, results are available through multiple tabs:

Model Summary: Shows model output, including estimates and diagnostics.
Convergence Diagnostics:
- R-hat, Geweke, Raftery-Lewis, Heidelberger-Welch tests
- Trace, density, ACF, NUTS energy, and ESS plots
Model Checking:
- Numerical: LOO, WAIC
- Graphical: Posterior predictive checks
Prior Sensitivity Analysis
Prediction:
- Display model-based small area estimates with uncertainty.
- Upload new data for out-of-sample predictions.
Update Model:
- Modify and refit the model with updated settings.
Save Output:
- Model Fit (RDS): Save fitted model object.
- Stan Code (TXT): Export Stan model code.
- MCMC Samples (CODA Format): Save posterior samples.
- Diagnostic Plots (PDF): Export diagnostic visualizations.

Example Workflow

Upload Dataset: Use .csv or select data from environment.
Explore Data: Summarize and visualize key variables.
Define Model: Set model structure, priors, and MCMC settings.
Prior Checking: Validate prior assumptions before sampling.
Fit Model: Run HBSAE using brms.
Review Results: Interpret summary and diagnostics.
Predict & Save: Generate estimates and export outputs.

Troubleshooting

If you encounter errors when launching the app:

Ensure all dependencies are installed manually:

install.packages(c("shiny", "shinyWidgets", "shinydashboard", "readxl", "DT"))

Reinstall hbsaems:

remove.packages("hbsaems")
install.packages("hbsaems")

Check the app directory:

system.file("shiny/sae_app", package = "hbsaems")

Conclusion

run_sae_app() provides an intuitive way to perform HBSAE modeling using a Shiny interface, making Bayesian small area estimation accessible without requiring in-depth coding knowledge. Users can define models, inspect results, and generate predictions interactively.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.