In FPRgroup/FPEMcountry: fpemlocal

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
knitr::opts_chunk$set(echo = TRUE)
library(FPEMlocal)
devtools::load_all()

Introduction

In this vignette we will fit FPET to multiple countries and aggregate the samples to obtain results for aggregate levels.

Table of Contents

1. Fit models
2. Aggregate samples
3. Calculate results

Fit models

Let's task ourselves with obtaining estimates of contraceptive use for the sub-region of Southern Africa.

divisions %>%
  dplyr::filter(name_sub_region == "Southern Africa") %>%
  dplyr::select(name_country) %>%
  unique()

First, we need fits for all countries in Southern Africa. Start by pulling a vector of country codes.

divs <- divisions %>%
  dplyr::filter(name_sub_region == "Southern Africa") %>%
  dplyr::filter(division_numeric_code %in% contraceptive_use$division_numeric_code) %>% # Filtering on the countries available in `contraceptive_use` because we are using this package data
  dplyr::pull(division_numeric_code)

Next, map the country codes to the function fit_fp_c. Note: Supply country codes as a list to be mapped.

first_year <- 1970
last_year <- 2030
union <- "Y"
fits <- purrr::pmap(list(division_numeric_code = divs %>% as.list(),
                         is_in_union = union,
                         first_year = first_year,
                         last_year = last_year
                         ),
                    fit_fp_c)

Aggregate samples

Within out list of fits we have more complex lists. We only need the samples from these fits. This function plucks the posterior samples and combines each array of samples along index 1 making one large array. The first dimension is for country codes. Use dimnames() on the resulting list to see how to index a country. This is the identical format of a posterior samples array from FPEMglobal.

posterior_samples <- pluck_abind_fp_c(fits)

Next, we need the population counts for these countries filtered by the time frame and union status.

## Get population data
population_data <- population_counts %>%
  dplyr::filter(division_numeric_code %in% divs) %>%
  dplyr::filter(is_in_union == union) %>%
  dplyr::filter(mid_year >= first_year) %>%
  dplyr::filter(mid_year <= last_year)

Now supply the posterior samples and population counts to the function aggregate_fp.

posterior_samples_aggregated <- aggregate_fp(posterior_samples =  posterior_samples,
                                             population_data = population_data)

Calculate results

Finally, supply the aggregated samples, the population data, and the first_year to calc_fp to calculate point estimates for family planning indicators.

results <- calc_fp(posterior_samples = posterior_samples_aggregated,
                   population_data = population_data,
                   first_year = first_year)

FPRgroup/FPEMcountry documentation built on April 24, 2023, 4:32 p.m.