targeting-validation.Rmd

---
title: "Targeting Validation"
author: "DIMA Americas"
date: " `r format(Sys.Date(),  '%d %B %Y')`"
always_allow_html: yes
output: unhcrdown::pptx_slides
---

```{r setup, include=FALSE}
#remotes::install_github('unhcr-web/unhcRstyle')
#install.packages("remotes")
#remotes::install_github("vidonne/unhcrthemes")
library(tidyverse)
library(tidymodels)
library(readxl)
library(unhcrthemes)
library(unhcrdown)

knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE)
knitr::opts_chunk$set(fig.width = 6, fig.height = 4, dpi = 150)
extrafont::loadfonts(quiet=TRUE)

data <- read_excel("data.xlsx")
```

# Score Distribution

```{r score-dist}
score_summary <- 
  data |> 
  summarize(across(c(democope, basicneeds), compose(enframe, summary))) |> 
  pivot_longer(everything()) |> 
  transmute(dim = name, quantile = value$name, value = as.numeric(value$value)) |> 
  filter(quantile == "Mean")

data |> 
  pivot_longer(-assisted, names_to = "dim", values_to = "score") |> 
  ggplot() +
  geom_histogram(aes(score), fill = "#0072BC", color = "white", binwidth = 5, boundary = 0) +
  geom_vline(aes(xintercept = value), data = score_summary) +
  geom_text(aes(x = value, y = Inf, label = str_c("Mean = ", round(value, 2))), 
            hjust = -.05, vjust = 1,
            data = score_summary) +
  facet_wrap(vars(dim)) +
  labs(x = "score", y = "# of cases",
       title = "Distribution of Scorecard Scores",
       subtitle = "Disaggregated by scoring dimension") +
  unhcrthemes::theme_unhcr(font_size = 11)
ggsave("file1.jpg")
```

# Score Association

```{r score-assoc}
data |> 
  pivot_longer(-assisted, names_to = "dim", values_to = "score") |> 
  ggplot(aes(score, as.numeric(as.logical(assisted)), color = dim)) +
  geom_smooth(method = "loess", se = FALSE) +
  geom_rug(sides = "b", position = "jitter") +
  scale_y_continuous(labels = scales::label_percent(), lim = c(0, 1)) +
  labs(x = "score", y ="% assisted", color = NULL,
       title = "Association between Scorecard Scores and Enumerator Decision",
       subtitle = "Disaggregated by scoring dimension") +
   unhcrthemes::theme_unhcr(font_size = 11) +
  theme(legend.position = "bottom")
ggsave("file2.jpg")
```

# Inclusion/Exclusion Error

```{r score-err}
data <- 
  data |> 
  mutate(amean = (democope+basicneeds)*(1/2),
         gmean = (democope*basicneeds)^(1/2),
         assisted = fct_rev(as_factor(assisted)))

athresh <- 
  data |> 
  roc_curve(assisted, amean) |> 
  arrange(desc(specificity+sensitivity)) |> 
  pluck(".threshold", 1)

gthresh <- 
  data |> 
  roc_curve(assisted, gmean) |> 
  arrange(desc(specificity+sensitivity)) |> 
  pluck(".threshold", 1)

data <- 
  data |> 
  mutate(apred = fct_rev(as_factor(amean > athresh)),
         gpred = fct_rev(as_factor(gmean > gthresh)))
```



```{r}
bind_rows(apred = data |> conf_mat(assisted, apred) |> summary(),
          gpred = data |> conf_mat(assisted, gpred) |> summary(),
          .id = "agg") |> 
  mutate(.estimate = if_else(.metric %in% c("sens", "spec"), 1-.estimate, .estimate)) |> 
  filter(.metric %in% c("sens", "spec", "accuracy", "kap")) |> 
  ggplot(aes(.metric, .estimate, group = agg)) +
  geom_col(aes(fill = agg), color = "white", position = position_dodge()) +
  geom_label(aes(label = scales::label_percent(.01)(.estimate)), position = position_dodge(.9)) +
  scale_x_discrete(labels = c(sens = "exclusion\nerror", spec = "inclusion\nerror", kap = "kappa")) +
  scale_y_continuous(labels = scales::label_percent()) +
  scale_fill_manual(values = c(apred = "black", gpred = "#0072BC"),
                    labels = c(apred = glue::glue("Arithmetic mean\n(thresh = {round(athresh, 2)})"), 
                               gpred = glue::glue("Geometric mean\n(thresh = {round(gthresh, 2)})"))) +
  coord_cartesian(clip = "off") +
  labs(x = "error measure", y = "% of cases", fill = NULL,
       title = "Targeting Error") +
   unhcrthemes::theme_unhcr(font_size = 11)
ggsave("file3.jpg")
unhcr_save(filename = "file3_unhcr.png", bg = "white")
```

```{r}
#rmarkdown::render("targeting-validation_GUAT1.Rmd") use it in console
```