wb_data_analysis.Rmd

---
title: "report"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message=FALSE, warning=FALSE)
```

```{r library}
library(tidyverse)
library(WDI)
library(vtree)
library(ggrepel)
```

```{r load_data}
# new_cache <- WDIcache()
# WDIsearch(string = "Access to electricity", field = "name", short = TRUE, cache = new_cache)
wb_data <- WDI(
  country = "all",
  indicator = c("electricity_access" = "EG.ELC.ACCS.ZS", 
                "national_pov" = "SI.POV.NAHC",
                "mort_att_water" = "SH.STA.WASH.P5",
                "tfr" = "SP.DYN.TFRT.IN",
                "child_perc" = "SP.POP.0014.TO.ZS"), 
  start = 1960,
  end = 2020,
  extra = TRUE,
#  cache = new_cache,
  latest = 100, # Integer indicating the number of most recent non-NA values to get.
  language = "en"
)
```


# Access to electricity
```{r}
# vtree ----
wb_data %>% 
  filter(year == 2019) %>% 
  mutate(less_than = ifelse(electricity_access < 20, country, "More than or equal to")) %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  vtree("region less_than", 
        showpct = FALSE,
        showcount = FALSE,
        summary="electricity_access \nAccess to electricity\n%mean% (%)",
        prune=list(less_than=c("More than or equal to")),
        labelvar=c(region = "Region"),
        showvarnames = FALSE,
        horiz = FALSE,
        title="**Access to electricity (% of population)** \n\n**World** \nNumber of countries"
  )
```

# Poverty vs access to electricity

There is negative relationship between access to electricity and poverty. For certain years South Africa and Ukraine stand out with high levels in both indicators.

```{r}
ggplot(data = wb_data, aes(x = electricity_access, y = national_pov)) +
  geom_point(aes(color = electricity_access > 75 & national_pov > 60)) +
  geom_label(aes(label = ifelse(electricity_access > 75 & national_pov > 60, 
                               paste0(country, "-", year), NA)),
            size = 2) +
  theme(legend.position = "none")
```

Have a look at recent years only.
```{r}
wb_data %>% 
  filter(year >= 2015) %>% 
ggplot(aes(x = electricity_access, y = national_pov)) +
  geom_point(aes(color = between(electricity_access, 75, 100) & between(national_pov, 40, 60))) +
  geom_label_repel(aes(label = ifelse(between(electricity_access, 75, 100) & between(national_pov, 40, 60), 
                               paste0(country, "-", year), NA)),
            size = 2) +
  theme(legend.position = "none")
```

Analyze Afghanistan, Honduras and Mexico. The increase in access to electricity in Afghanistan is very interesting. Honduras on the other hand sometimes experience decreases in access to electricity.

```{r}
wb_data %>% 
  filter(country %in% c("Afghanistan", "Honduras", "Mexico")) %>% 
ggplot(aes(x = electricity_access, y = national_pov)) +
  geom_point() +
  geom_label_repel(aes(label = paste0(country, "-", year)), 
                   size = 2) +
  theme(legend.position = "none")
```

Countries having electricity access less than 50 percent in any year
```{r}
elec_less_than_50 <- 
  wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% # filter out aggregates
  filter(electricity_access < 50) %>% 
  select(country) %>%
  unique() %>% 
  unlist(use.names = FALSE)
```

In the data, there are a total of `r length(elec_less_than_50)` countries that had less than 50 percent electricity access.

Early improvers. Increase to over 80 percent before 2010.
```{r}
wb_data %>% 
  filter(country %in% elec_less_than_50) %>% 
ggplot(aes(x = year, y = electricity_access)) +
  geom_point() +
  geom_label_repel(aes(label = ifelse(electricity_access > 80 & year < 2010, 
                                      paste0(country, "-", year), NA)))
```

These countries are Indonesia and Morocco. Let's see how they evolved.

```{r}
wb_data %>% 
  filter(country %in% c("Indonesia", "Morocco")) %>% 
ggplot(aes(x = year, y = electricity_access)) +
  geom_point() +
  geom_label_repel(aes(label = paste0(country, "-", round(electricity_access, 1))))
```

This is not a good way to visualize. Let's use geom_line. Sudden increase and decreases could be related to data quality issues.

```{r}
wb_data %>% 
  filter(country %in% c("Indonesia", "Morocco")) %>% 
ggplot(aes(x = year, y = electricity_access, color = country)) +
  geom_line() 
```

All countries with values below 50 percent.
```{r}
wb_data %>% 
  filter(country %in% elec_less_than_50) %>% 
ggplot(aes(x = year, y = electricity_access, color = country)) +
  geom_line() +
  theme(legend.position = "none")
```

Sub-Saharan Africa
All countries with values below 50 percent.
```{r}
wb_data %>% 
  filter(country %in% elec_less_than_50) %>% 
  filter(region == "Sub-Saharan Africa") %>% 
ggplot(aes(x = year, y = electricity_access, color = country)) +
  geom_line() +
  scale_x_continuous(limits = c(1990, 2020)) +
  theme(legend.position = "none")
```

Which increased over 60 percent.

```{r}
improved_sub_sahara_c <- 
wb_data %>% 
  filter(country %in% elec_less_than_50) %>% 
  filter(region == "Sub-Saharan Africa") %>% 
  filter(electricity_access >= 60) %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```

```{r}
wb_data %>% 
  filter(country %in% improved_sub_sahara_c) %>% 
  mutate(label = if_else(year == max(year), as.character(country), NA_character_)) %>% 
ggplot(aes(x = year, y = electricity_access, color = country)) +
  geom_line() +
  geom_label_repel(aes(label = label)) +
  scale_x_continuous(limits = c(1990, 2020)) +
  theme(legend.position = "bottom")
```

Compare minimum and maximums.

```{r}
most_improve_5_c <- 
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates" & !is.na(electricity_access)) %>%
  group_by(country) %>% 
  summarise(min_elec = min(electricity_access), max_elec = max(electricity_access)) %>% 
  mutate(range_elec = round(max_elec - min_elec, 1)) %>% 
  arrange(desc(range_elec)) %>% 
  head(5) %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```
```{r}
wb_data %>% 
  filter(country %in% most_improve_5_c) %>% 
ggplot(aes(x = year, y = electricity_access, color = country)) +
  geom_line()
```

# Mortality rate attributed to unsafe water

```{r}
wb_data %>% 
  ggplot(aes(x = year, y = mort_att_water)) +
  geom_point()
```

```{r}
wb_data %>% 
  filter(!is.na(mort_att_water)) %>% 
  ggplot(aes(x = country, y = mort_att_water)) +
  geom_col()
```

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  filter(!is.na(mort_att_water) & mort_att_water >= 25) %>% 
  ggplot(aes(x = country, y = mort_att_water, fill = region)) +
  geom_col() +
  geom_label_repel(aes(label = ifelse(region == "Middle East & North Africa" | mort_att_water >= 50, 
                                      country, NA))) +
  theme(legend.position = "bottom")
```

Infant mortality

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  ggplot(aes(x = year, y = tfr)) +
  geom_point()
```

Countries with fertility over 7.5 at any stage in the data.

```{r}
high_fert_c <- 
wb_data %>% 
  filter(tfr >= 7.5) %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```

Look at min, max and the difference between for these countries.

```{r}
max_diff_c <- 
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  filter(country %in% high_fert_c & !is.na(tfr)) %>% 
  group_by(country) %>% 
  summarise(max_fert = max(tfr),
            min_fert = min(tfr)) %>% 
  mutate(diff_max_min = max_fert - min_fert) %>% 
  arrange(desc(diff_max_min)) %>% 
  head() %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  filter(country %in% max_diff_c) %>% 
ggplot(aes(x = year, y = tfr, color = country)) +
  geom_line()
```

```{r}
min_diff_c <- 
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  filter(country %in% high_fert_c & !is.na(tfr)) %>% 
  group_by(country) %>% 
  summarise(max_fert = max(tfr),
            min_fert = min(tfr)) %>% 
  mutate(diff_max_min = max_fert - min_fert) %>% 
  arrange(desc(diff_max_min)) %>% 
  tail(3) %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```


```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  filter(country %in% min_diff_c) %>% 
ggplot(aes(x = year, y = tfr, color = country)) +
  geom_line()
```

Max min difference for all countries

```{r}
max_diff_c_all <- 
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  filter(!is.na(tfr)) %>% 
  group_by(country) %>% 
  summarise(max_fert = max(tfr),
            min_fert = min(tfr)) %>% 
  mutate(diff_max_min = max_fert - min_fert,
         perc_diff_max_min = (max_fert - min_fert) / max_fert * 100) %>% 
  arrange(desc(perc_diff_max_min)) %>% 
  head() %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  filter(country %in% max_diff_c_all) %>% 
ggplot(aes(x = year, y = tfr, color = country)) +
  geom_line()
```



```{r}
min_diff_c_all <- 
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>% 
  filter(!is.na(tfr)) %>% 
  group_by(country) %>% 
  summarise(max_fert = max(tfr),
            min_fert = min(tfr)) %>% 
  mutate(diff_max_min = max_fert - min_fert,
         perc_diff_max_min = (max_fert - min_fert) / max_fert * 100) %>% 
  arrange(desc(perc_diff_max_min)) %>% 
  tail() %>% 
  select(country) %>% 
  unique() %>% 
  unlist(use.names = FALSE)
```

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  filter(country %in% c(min_diff_c_all, max_diff_c_all)) %>% 
ggplot(aes(x = year, y = tfr, color = country)) +
  geom_line()
```

Can regions be visualised?
```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  ggplot(aes(x = year, y = tfr, color = region)) + 
  geom_point()
```

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  ggplot(aes(x = year, y = tfr, color = income)) + 
  geom_point()
```

Lower middle income c's. Two countries with low levels.
```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  filter(income == "Lower middle income") %>% 
  ggplot(aes(x = year, y = tfr, color = country)) + 
  geom_point() +
  geom_label_repel(aes(label = ifelse(year == 1960 & tfr < 4, paste0(country), NA)))
```

High middle income c's. Two countries with low levels.
```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  filter(income == "High income") %>% 
  ggplot(aes(x = year, y = tfr, color = country)) + 
  geom_point() +
  geom_label_repel(aes(label = ifelse((year == 2019 & tfr > 2.5) | (year == 1960 & tfr > 7), paste0(country, "-", round(tfr, 1)), NA))) +
  theme(legend.position =  "none")
```

# Children percentage

```{r}
wb_data %>% 
  filter(!is.na(region) & region != "Aggregates") %>%
  ggplot(aes(x = year, y = child_perc, color = region)) + 
  geom_point()
```

Europe and Central Asia

```{r}
wb_data %>% 
  filter(region == "Europe & Central Asia") %>%
  ggplot(aes(x = year, y = child_perc, color = country)) + 
  geom_point() +
  geom_label_repel(aes(label = ifelse(year == 2020 & child_perc > 25, country, NA))) +
  theme(legend.position =  "none")
```

```{r}
wb_data %>% 
  filter(region == "Sub-Saharan Africa") %>%
  ggplot(aes(x = year, y = child_perc, color = country)) + 
  geom_point() +
  geom_label_repel(aes(label = ifelse(year == 2020 & (child_perc < 35 | child_perc > 47), country, NA))) +
  theme(legend.position =  "none")
```