vignettes/in_union_women_from_custom_data.md

In FPRgroup/FPEMcountry: fpemlocal

Estimating family planning indicators for married women with custom user data ================

library(fpemlocal)

Introduction

In this vignette we obtain estimates for married women with custom user data.

1. Fit a one country model fit_fp_c
2. Calculate point estimates for indicators calc_fp_c
3. Plot the point estimates against the survey data plot_fp_c

1. Fit a one country model

Fit the one-country family planning estimation model with the function fit_fp_c. Supply the file path of the .csv file containing your country survey data to the argument surveydata_filepath. Next supply the UNPD country code known as the division_numeric_code. See ??divisions for UNPD country codes and other divisions. Specify the model of your choice with the argument is_in_union. There are two versions of this model, one for in-union women and another for not-in-union women denoted "Y" and "N" respectively. Lastly, specify the years of estimates to be returned. Note: The function will not filter the survey data based on these years. All years of available survey data will be used.

fit <- fit_fp_c(
  surveydata_filepath = "data-raw/manuscript_example_data/afghanistan_4_married_example.csv",
  division_numeric_code = 4,
  is_in_union = "Y",
  first_year = 1970,
  last_year = 2030
)

## 
## -- Column specification --------------------------------------------------------
## cols(
##   .default = col_double(),
##   is_in_union = col_character(),
##   age_range = col_character(),
##   data_series_type = col_character(),
##   group_type_relative_to_baseline = col_character(),
##   unmet_need_modern = col_logical(),
##   is_pertaining_to_methods_used_since_last_pregnancy = col_character(),
##   pertaining_to_methods_used_since_last_pregnancy_reason = col_logical(),
##   has_geographical_region_bias = col_character(),
##   geographical_region_bias_reason = col_character(),
##   has_non_pregnant_and_other_positive_biases = col_character(),
##   non_pregnant_and_other_positive_biases_reason = col_logical(),
##   age_group_bias = col_character(),
##   modern_method_bias = col_character(),
##   has_traditional_method_bias = col_character(),
##   traditional_method_bias_reason = col_logical(),
##   has_absence_of_probing_questions_bias = col_character(),
##   record_id = col_character()
## )
## i Use `spec()` for the full column specifications.

2. Calculate point estimates for indicators

Calculate point estimates for family planning indicators with the function calc_fp_c. Supply the fit object from fit_fp_c to the argument fit. Read in your population count dataset. Then supply the dataset to the argument population_data.

population_data <- read.csv("data-raw/manuscript_example_data/afghanistan_4_married_popdata_example.csv")
results <- calc_fp_c(fit = fit,
                     population_data = population_data)

A set of results here consist of the following family planning indicators

results$Y %>% names

##  [1] "contraceptive_use_any"                     
##  [2] "contraceptive_use_modern"                  
##  [3] "contraceptive_use_traditional"             
##  [4] "non_use"                                   
##  [5] "unmet_need_any"                            
##  [6] "unmet_need_modern"                         
##  [7] "demand"                                    
##  [8] "demand_modern"                             
##  [9] "demand_satisfied"                          
## [10] "demand_satisfied_modern"                   
## [11] "no_need"                                   
## [12] "contraceptive_use_any_population_counts"   
## [13] "contraceptive_use_modern_population_counts"
## [14] "traditional_cpr_population_counts"         
## [15] "non_use_population_counts"                 
## [16] "unmet_need_population_counts"              
## [17] "unmet_need_modern_population_counts"       
## [18] "demand_modern_population_counts"           
## [19] "demand_population_counts"                  
## [20] "demand_satisfied_population_counts"        
## [21] "demand_satisfied_modern_population_counts" 
## [22] "no_need_population_counts"

The point estimates for each indicator are long-format tibbles. Let’s take a look at the tibble for the indicator contraceptive_use_modern

results$Y$contraceptive_use_modern

## # A tibble: 488 x 3
##     year percentile  value
##    <int> <chr>       <dbl>
##  1  1970 mean       0.0122
##  2  1971 mean       0.0130
##  3  1972 mean       0.0139
##  4  1973 mean       0.0149
##  5  1974 mean       0.0160
##  6  1975 mean       0.0173
##  7  1976 mean       0.0187
##  8  1977 mean       0.0201
##  9  1978 mean       0.0217
## 10  1979 mean       0.0234
## # ... with 478 more rows

3. Plot estimates and survey data

fpemlocal also includes a function named plot_fp_c to plot the calculated point estimates against the survey data. The arguments to this function are, the fit object from step 1, the results from step 2, and a vector of indicator names. The vector of indicator names corresponds to the names which appear in the results from step 2.

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

## $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.