README.md

In FPRgroup/FPEMcountry: fpemlocal

fpemlocal

Installation

The package can be installed by cloning and using devtools::install(). The source code for vignettes can be found in /vignettes. Below is a brief introduction.

Introduction

The fpemlocal package is the one-country implementation of FPEM (family planning estimation model) designed with tidyverse philosophy. The model in this package uses global model results from the package FPEMglobal to aid in the estimation of country level family planning indicators. fpemlocal comes equiped with survey data, country unit data, and country population count data, to produce one-country runs. Running FPEM is divided into three main functions.

1. Fit a one country model fit_fp_c
2. Calculate family planning indicators clac_fp_c
3. Plot the family planning estimates against the survey data plot_fp_c

These three functions make running one country FPEM straightforward, while retaining enough division to carry out a variety of developer and client tasks. In this document we will cover the typical use of these three functions.

To start a run we need to know the country code for the country of interest. Our package contains country codes and other country units in the dataset divisions.

divisions %>% head

## # A tibble: 6 x 13
##   division_numeri~ name_country name_region name_sub_region region_numeric_~
##              <dbl> <chr>        <chr>       <chr>                      <dbl>
## 1                4 Afghanistan  Asia        South-Central ~              935
## 2                8 Albania      Europe      Southern Europe              908
## 3               12 Algeria      Africa      Northern Africa              903
## 4               16 American Sa~ Oceania     Polynesia                    909
## 5               20 Andorra      Europe      Southern Europe              908
## 6               24 Angola       Africa      Middle Africa                903
## # ... with 8 more variables: sub_region_numeric_code <dbl>,
## #   is_developed_region <chr>, is_less_developed_region <chr>,
## #   is_least_developed_country <chr>, is_in_sub_saharan_africa <chr>,
## #   is_unmarried_sexual_activity <chr>, is_low_population <chr>,
## #   is_fp2020 <chr>

Our package data sets are tibbles. The country codes used by our package, known as division_numeric_code, are found in this data. In our example we will execute a one-country run for Afghanistan, code 4. Survey data is available in the dataset contraceptive_use. See ??contraceptive_use for a detailed description of this dataset.

1. Fit a one country model

fit_fp_c is a wrapper function to fit the one-country implementation of the family planning estimation model. There are two versions of this model, one for in-union and another for not-in-union women which can be specified with the argument is_in_union. These are denoted "Y" and "N" respectively. The first_year and last_year arguments determine the years of estimates exported from the run. Regardless of these arguments, the function will use all years in which data is available for estimation. When a survey file is not provided, as in this example, the function uses default package contraceptive_use. The user may also supply optional services statistics.

fit <- fit_fp_c(
  is_in_union = "Y",
  division_numeric_code = 4,
  first_year = 1970,
  last_year = 2030,
  diagnostic = TRUE
)

2. Calculate point estimates for indicators

fpet_calculate_indicators is a wrapper function for calculating point estimates and confidence intervals. By default the function uses package population data (See population_counts) in order to calculate family planning indicators. Custom population count data may be supplied (See ??fpet_get_results).

results <- calc_fp_c(fit)

3. Plot the point estimates against the survey data

plot_fp_c plots the results of the model against the survey data. The user supplies the objects exported from fit_fp_c and fpet_calculate_indicators as well as indicators of interest. Indicators of interest are supplied to the argument indicators. The argument compare_to_global adds point estimate and 95% credible interval from the UNPD global model (See global_estimates). The global model estimates are plotted using dotted lines. Since we are only using the default data from UNPD the estimates from our model should align with the UNPD estimates.

plot_fp_c(
  fit,
  results,
  indicators = c(
    "unmet_need_any",
    "contraceptive_use_modern",
    "contraceptive_use_traditional",
    "contraceptive_use_any"
    ),
  compare_to_global = FALSE
)

## $Y
## $Y$unmet_need_any

## 
## $Y$contraceptive_use_modern

## 
## $Y$contraceptive_use_traditional

## 
## $Y$contraceptive_use_any

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