Package 'bayescoveragedeploy'

Title: Precompiled Stan Models for Deployment of Bayesian Coverage Model
Description: Provides precompiled Stan models for fitting local country-level Bayesian hierarchical transition models for health coverage indicators. This package uses the rstan backend with precompiled models for users without C++ compilers. All core functionality is provided by the bayescoveragemodel package.
Authors: Leontine Alkema [aut, cre]
Maintainer: Leontine Alkema <[email protected]>
License: MIT + file LICENSE
Version: 0.21
Built: 2026-06-10 03:00:20 UTC
Source: https://github.com/AlkemaLab/bayescoveragedeploy

Help Index


The 'bayescoveragedeploy' package.

Description

Provides precompiled Stan models for fitting local country-level Bayesian hierarchical transition models for health coverage indicators (ANC4, institutional delivery, vaccination). This package uses the rstan backend with precompiled models for users without C++ compilers. All core functionality is provided by the bayescoveragemodel package.

Details

This package contains precompiled Stan models for the following variants:

  • fpem - Basic transition model

  • fpem_routine - Model with routine data

  • fpem_aggregates - Model with subnational aggregates

  • fpem_routine_aggregates - Model with routine data and aggregates

The main user-facing function is fit_local_model.

Author(s)

Maintainer: Leontine Alkema [email protected]

References

Stan Development Team (NA). RStan: the R interface to Stan. R package version 2.32.7. https://mc-stan.org

See Also

Useful links:


Fit local country model with precompiled Stan models

Description

This function provides a simplified interface for fitting country-specific models using precompiled Stan models. All data processing and model logic is delegated to bayescoveragemodel::fit_model().

Usage

fit_local_model(
  survey_df,
  iso_select,
  indicator = "anc4",
  routine_df = NULL,
  chains = 4,
  seed = 1234,
  refresh = 10,
  iter_sampling = 300,
  iter_warmup = 150,
  adapt_delta = 0.95,
  max_treedepth = 14,
  ...
)

Arguments

survey_df

Survey data (processed via bayescoveragemodel::process_data)

iso_select

ISO code for the country

indicator

Indicator name ("anc4", "ideliv", "vdpt", etc.)

routine_df

Optional tibble with routine data

chains

Number of MCMC chains (default 4)

seed

Random seed (default 1234)

refresh

Progress update frequency (default 10)

iter_sampling

Number of sampling iterations (default 200)

iter_warmup

Number of warmup iterations (default 150)

adapt_delta

Target acceptance rate (default 0.9)

max_treedepth

Maximum tree depth (default 14)

...

Additional arguments passed to bayescoveragemodel::fit_model()

Value

Model fit object (same structure as bayescoveragemodel::fit_model)

Examples

## Not run: 
# ===== Quick Example =====
library(haven)
dat0 <- read_dta("data_raw/ICEH_national.dta")
regions_dat <- readr::read_csv("data_raw/regions_updated.csv")

# Process data
data <- bayescoveragemodel::process_data(
  dat = dat0,
  regions_dat = regions_dat,
  indicator = "anc4"
)

# Fit model (fast test)
fit <- fit_local_model(
  survey_df = data,
  iso_select = "KEN",
  indicator = "anc4",
  chains = 1,
  iter_sampling = 5,
  iter_warmup = 5
)

# Plot results
bayescoveragemodel::plot_estimates_local_all(fit)

# ===== Production Example =====
fit_prod <- fit_local_model(
  survey_df = data,
  iso_select = "KEN",
  indicator = "anc4",
  chains = 4,
  iter_sampling = 200,
  iter_warmup = 150,
  seed = 123
)

# ===== More Examples =====
# See BayesCoverage app!

## End(Not run)