Experimenting with distance as a concept for SDG achievement

Exploring how to conceptualize and visualize distance from Sustainable Development Goal targets
R
data-viz
tidyverse
international-development
cluster-analysis
mapping
Author

Mickey Rafa

Published

April 15, 2019

Introduction

Purpose

The United Nations created the Sustainable Development Goals (SDGs) to set an ambitious global development agenda to work toward by 2030. A continental development organization in Africa asked, how could we think about SDG 3 (the health goals) in a holistic and aggregate sense?

In this analysis, I explore measuring a country’s distance to target achievement. I build a composite score of standardized distance to SDG 3 achievement, and find that Nigeria is furthest from achieving SDG 3 of all African countries.

Dataset

This analysis uses results from the International Futures global forecasting model and its Current Path scenario. The results span many of the measurable targets of SDG3, including:

  • Maternal mortality ratio
  • Neonatal mortality ratio
  • Under-5 mortality rate
  • Non-communicable disease death rate
  • Traffic accident death rate
  • AIDS death rate
  • Malaria death rate

Setup

library(tidyverse)
── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.1     ✔ tibble    3.2.1
✔ lubridate 1.9.3     ✔ tidyr     1.3.1
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
#code for the targets of each health-related goal
sdg_targets <- data.frame(
  variable = c("MATMORTRATIO", "NEONATMOR", "CHILDDTHR", "NonCommun", "TrafficAcc", "AIDS", "Malaria"),
  country = rep("TARGET", 7),
  `2030` = c(70, 12, 25, 2/3, 0.5, 0.001, 0.001)
) %>% 
  rename("2030" = "X2030")

(df <- read_csv('.//data/IFs_results_02apr2019.csv') %>% 
   filter(scenario == '00CP_adj_ADB') %>% 
   select(1:5, `2015`, `2020`, `2030`, `2063`))
Rows: 1458 Columns: 111
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr  (5): variable, country, unit, dim, scenario
dbl (89): 1960, 1961, 1962, 1963, 1964, 1965, 1966, 1967, 1968, 1969, 1970, ...
num (17): 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, ...

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
# A tibble: 486 × 9
   variable  country  unit       dim   scenario     `2015` `2020` `2030` `2063`
   <chr>     <chr>    <chr>      <chr> <chr>         <dbl>  <dbl>  <dbl>  <dbl>
 1 DeathRate Algeria  TrafficAcc Total 00CP_adj_ADB 0.268  0.271  0.284   0.337
 2 DeathRate Algeria  AIDS       Total 00CP_adj_ADB 0.005  0.008  0.005   0.001
 3 DeathRate Algeria  Malaria    Total 00CP_adj_ADB 0.0002 0.0002 0.0001  0    
 4 DeathRate Angola   TrafficAcc Total 00CP_adj_ADB 0.386  0.37   0.402   0.525
 5 DeathRate Angola   AIDS       Total 00CP_adj_ADB 0.395  0.314  0.198   0.021
 6 DeathRate Angola   Malaria    Total 00CP_adj_ADB 0.52   0.512  0.375   0.052
 7 DeathRate Benin    TrafficAcc Total 00CP_adj_ADB 0.213  0.219  0.229   0.387
 8 DeathRate Benin    AIDS       Total 00CP_adj_ADB 0.274  0.203  0.111   0.007
 9 DeathRate Benin    Malaria    Total 00CP_adj_ADB 1.21   0.808  0.424   0.083
10 DeathRate Botswana TrafficAcc Total 00CP_adj_ADB 0.174  0.177  0.183   0.192
# ℹ 476 more rows
(df_scaled_dist_results <- df %>% 
  filter(variable %in% c('MATMORTRATIO', 'NEONATMOR', 'CHILDDTHR') | 
         unit %in% c('AIDS', 'Malaria')) %>% 
  mutate(variable = ifelse(unit %in% c('AIDS', 'Malaria'), as.character(unit), as.character(variable))) %>% 
  select(variable, country, `2030`) %>% 
  spread(variable, `2030`) %>% 
  #add targets as row by reshaping the targets dataframe
  bind_rows(., 
            sdg_targets %>%
              spread(variable, `2030`)) %>% 
  mutate_at(vars(AIDS:NEONATMOR), funs(scale(.))) %>% 
  select(-c(7:8)))
# A tibble: 55 × 6
   country     AIDS[,1] CHILDDTHR[,1] Malaria[,1] MATMORTRATIO[,1] NEONATMOR[,1]
   <chr>          <dbl>         <dbl>       <dbl>            <dbl>         <dbl>
 1 Algeria       -0.812      -1.04        -0.785            -1.04        -0.440 
 2 Angola        -0.340       1.22         0.363             0.417        3.07  
 3 Benin         -0.553       0.475        0.513            -0.117        0.0158
 4 Botswana       1.29       -0.708       -0.766            -1.20        -0.126 
 5 Burkina Fa…   -0.648       0.386        1.16             -0.227       -0.182 
 6 Burundi       -0.575       0.00822     -0.0256            1.11        -0.212 
 7 Cameroon       0.432       0.805        1.29              0.683        0.376 
 8 Cape Verde    -0.294      -1.27        -0.784            -1.46        -0.971 
 9 Central AfR    0.569       1.73         0.501             1.63         1.32  
10 Chad          -0.558       2.78         0.648             1.78         2.14  
# ℹ 45 more rows
(df_scaled_dist_results_rel_targets <- df %>% 
  filter(unit %in% c('TrafficAcc', 'NonCommun')) %>% 
  select(unit, country, `2015`, `2020`, `2030`) %>% 
  rename('variable' = 'unit') %>% 
  mutate(ratio_ = ifelse(variable == 'TrafficAcc', `2020` / `2015`, `2030` / `2015`)) %>% 
  select(-c(3:5)) %>% 
  spread(variable, ratio_) %>% 
  bind_rows(., 
            #add targets as row by reshaping the targets dataframe
            sdg_targets %>%
              filter(variable %in% c('TrafficAcc', 'NonCommun')) %>% 
              spread(variable, `2030`)) %>% 
  mutate_at(vars(NonCommun:TrafficAcc), funs(scale(.))))
Warning: `funs()` was deprecated in dplyr 0.8.0.
ℹ Please use a list of either functions or lambdas:

# Simple named list: list(mean = mean, median = median)

# Auto named with `tibble::lst()`: tibble::lst(mean, median)

# Using lambdas list(~ mean(., trim = .2), ~ median(., na.rm = TRUE))
# A tibble: 55 × 3
   country      NonCommun[,1] TrafficAcc[,1]
   <chr>                <dbl>          <dbl>
 1 Algeria              1.52          0.0612
 2 Angola              -0.304        -0.570 
 3 Benin               -0.518         0.265 
 4 Botswana             1.66          0.134 
 5 Burkina Faso        -0.640         0.306 
 6 Burundi             -0.928        -1.03  
 7 Cameroon            -0.761         0.319 
 8 Cape Verde          -0.388         0.539 
 9 Central AfR         -0.497        -0.0730
10 Chad                -0.974        -0.452 
# ℹ 45 more rows
my.theme.minimal <- theme(
  plot.title = element_text(color="black", face="plain", size=10, hjust=0), 
  plot.subtitle = element_text(color="black", size=15, hjust=0), 
  axis.title = element_text(color="black", face="plain", size=14), 
  axis.text.y = element_text(size=13), 
  axis.text.x = element_text(angle = 90, hjust = 1, size=16), 
  axis.line = element_line(linewidth = .5, color = "black"),
  plot.caption = element_text(color="black", size=13), 
  panel.background =  element_blank(), 
  panel.border = element_blank(), 
  legend.title=element_blank(), 
  legend.text = element_text(color="black", size=14, hjust=0),
  legend.spacing.x = unit(.5, 'cm'), 
  legend.position = 'bottom',
  strip.text = element_text(color="black", face="plain", size=16),
  strip.background = element_rect(fill = "white"))

Setting order by distance from achievement

# A tibble: 54 × 2
   country               avg_val
   <chr>                   <dbl>
 1 Libya                  -0.703
 2 Sao Tome and Principe  -0.677
 3 Cape Verde             -0.661
 4 Seychelles             -0.635
 5 Egypt                  -0.630
 6 Tunisia                -0.552
 7 Senegal                -0.453
 8 Niger                  -0.387
 9 Eritrea                -0.373
10 Algeria                -0.363
# ℹ 44 more rows
(dist_plot_df <- df_scaled_dist_results %>% 
  filter(country != 'TARGET') %>% 
  left_join(x=.,
            y=df_scaled_dist_results_rel_targets,
            by='country') %>% 
  left_join(x=.,
            y=cntry_order,
            by='country') %>% 
  arrange(-avg_val) %>% 
  gather(var, val, 2:8))
Warning: attributes are not identical across measure variables; they will be
dropped
# A tibble: 378 × 4
   country     avg_val var        val
   <chr>         <dbl> <chr>    <dbl>
 1 Nigeria       1.15  AIDS  -0.00515
 2 Somalia       0.922 AIDS  -0.711  
 3 Equa Guinea   0.843 AIDS   0.902  
 4 Chad          0.766 AIDS  -0.558  
 5 Central AfR   0.741 AIDS   0.569  
 6 SierraLeo     0.610 AIDS  -0.372  
 7 Lesotho       0.610 AIDS   3.90   
 8 Angola        0.551 AIDS  -0.340  
 9 Sudan South   0.527 AIDS   1.02   
10 Cameroon      0.449 AIDS   0.432  
# ℹ 368 more rows
cntry_order$country <- as.factor(cntry_order$country)
(df_target_scaled_vals <- df_scaled_dist_results %>% 
  filter(country == 'TARGET') %>% 
  left_join(x=.,
            y=df_scaled_dist_results_rel_targets %>% filter(country == 'TARGET'),
            by='country') %>% 
  gather(var, target, 2:8) %>% 
  select(2:3))
Warning: attributes are not identical across measure variables; they will be
dropped
# A tibble: 7 × 2
  var          target
  <chr>         <dbl>
1 AIDS         -0.821
2 CHILDDTHR    -0.868
3 Malaria      -0.782
4 MATMORTRATIO -1.08 
5 NEONATMOR    -0.516
6 NonCommun    -3.05 
7 TrafficAcc   -6.07 
cbPalette <- c("#999999", "#E69F00", "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#D55E00", "#CC79A7")

targ_cols <- c('AIDS' = "#E69F00", 
               'CHILDDTHR' = "#56B4E9",
               'Malaria' = "#009E73",
               'MATMORTRATIO' = "#F0E442",
               'NEONATMOR' = "#0072B2",
               'NonCommun' = "#D55E00",
               'TrafficAcc' = "#CC79A7")
(dist_plot_df_2 <- dist_plot_df %>% 
  left_join(x=.,
            y=df_target_scaled_vals,
            by='var'))
# A tibble: 378 × 5
   country     avg_val var        val target
   <chr>         <dbl> <chr>    <dbl>  <dbl>
 1 Nigeria       1.15  AIDS  -0.00515 -0.821
 2 Somalia       0.922 AIDS  -0.711   -0.821
 3 Equa Guinea   0.843 AIDS   0.902   -0.821
 4 Chad          0.766 AIDS  -0.558   -0.821
 5 Central AfR   0.741 AIDS   0.569   -0.821
 6 SierraLeo     0.610 AIDS  -0.372   -0.821
 7 Lesotho       0.610 AIDS   3.90    -0.821
 8 Angola        0.551 AIDS  -0.340   -0.821
 9 Sudan South   0.527 AIDS   1.02    -0.821
10 Cameroon      0.449 AIDS   0.432   -0.821
# ℹ 368 more rows
(seg_length <- dist_plot_df_2 %>% 
  mutate(achieve = ifelse(val <= target, 1, 0)) %>% 
  filter(achieve == 0) %>% 
  group_by(country) %>% 
  summarise(min_val = min(val),
            max_val = max(val)))
# A tibble: 54 × 3
   country      min_val max_val
   <chr>          <dbl>   <dbl>
 1 Algeria       -1.04    1.52 
 2 Angola        -0.570   3.07 
 3 Benin         -0.553   0.513
 4 Botswana      -0.766   1.66 
 5 Burkina Faso  -0.648   1.16 
 6 Burundi       -1.03    1.11 
 7 Cameroon      -0.761   1.29 
 8 Cape Verde    -0.388   0.539
 9 Central AfR   -0.497   1.73 
10 Chad          -0.974   2.78 
# ℹ 44 more rows
dist_plot_df_2 %>% 
  mutate(achieve = ifelse(val <= target, 1, 0)) %>% 
  filter(achieve == 0) %>% 
  ggplot(.,
         aes(x=val, 
             y=factor(country, levels=cntry_order$country))) + 
  geom_point(aes(color=var), 
             #height=.25, width=0,
             alpha=.5, size=8) + 
  geom_segment(data=seg_length,
               aes(x=min_val, xend=max_val, yend=country),
               color='gray60', linewidth=1) + 
  labs(x='',
       y='') + 
  my.theme.minimal + 
  theme(legend.position = 'top') + 
  scale_color_manual(values = targ_cols,
                     labels = c('MATMORTRATIO' = 'Maternal Mortality',
                               'NEONATMOR' = 'Neonatal Mortality',
                               'CHILDDTHR' = '< 5 Mortality',
                               'NonCommun' = 'NCDs',
                               'TrafficAcc' = 'Traffic Mortality')) + 
  scale_y_discrete(labels = function(x) str_wrap(x, width = 10)) + 
  scale_x_continuous(position = 'top')