2020_09_01_crop_yields.Rmd

---
title: "TidyTemplate"
date: 2020-09-01
output: html_output
---

```{r setup, include=FALSE}

knitr::opts_chunk$set(echo = TRUE)

library(tidyverse)
library(tidytuesdayR)
library(scales)
library(janitor)
theme_set(theme_light())

```

# Load the weekly Data

Dowload the weekly data and make available in the `tt` object.

```{r Load}
tt <- tt_load("2020-09-01")
```

```{r}
tt$key_crop_yields %>%
  View()

yields <- tt$key_crop_yields %>%
  clean_names() %>%
  rename_all(str_remove, "_tonnes.*")

arable_land <- tt$arable_land_pin %>%
  clean_names() %>%
  rename(arable_land_needed = 4)

arable_land %>%
  filter(entity == "Afghanistan") %>%
  ggplot(aes(year, arable_land_needed)) +
  geom_line()

fertilizer <- tt$cereal_crop_yield_vs_fertilizer_application %>%
  janitor::clean_names() %>%
  rename(yield = 4,
         fertilizer_use = 5)
```

```{r}
yields_tidy <- yields %>%
  pivot_longer(wheat:bananas, names_to = "crop", values_to = "yield") %>%
  filter(!is.na(yield)) %>%
  mutate(crop = str_replace_all(crop, "_", " "),
         crop = str_to_title(crop))

yields_tidy %>%
  write_rds("crop-yields-shiny/yields_tidy.rds")
```

```{r}
yields_tidy %>%
  filter(code == "USA") %>%
  mutate(crop = fct_reorder(crop, -yield)) %>%
  ggplot(aes(year, yield)) +
  geom_line() +
  facet_wrap(~ crop)

yields_tidy %>%
  filter(code == "USA") %>%
  mutate(crop = fct_reorder(crop, -yield)) %>%
  ggplot(aes(year, yield, color = crop)) +
  geom_line() +
  labs(x = "Year",
       y = "Yield (tonnes per hectare)",
       title = "Crop yields in the US over time",
       color = "Crop")

yields_tidy %>%
  filter(entity == "India") %>%
  mutate(crop = fct_reorder(crop, -yield)) %>%
  ggplot(aes(year, yield, color = crop)) +
  geom_line() +
  labs(x = "Year",
       y = "Yield (tonnes per hectare)",
       title = "Crop yields in the US over time",
       color = "Crop")
```

```{r}
yields_tidy %>%
  filter(crop == "Wheat") %>%
  add_count(entity) %>%
  filter(n == max(n)) %>%
  filter(entity %in% sample(unique(entity), 25)) %>%
  ggplot(aes(year, yield)) +
  geom_line() +
  facet_wrap(~ entity)

crop_yields_50_years <- yields_tidy %>%
  arrange(entity, year) %>%
  filter(year >= 1968) %>%
  group_by(entity, code, crop) %>%
  summarize(year_start = min(year),
            year_end = max(year),
            yield_start = first(yield),
            yield_end = last(yield)) %>%
  ungroup() %>%
  filter(year_start == 1968) %>%
  mutate(yield_ratio = yield_end / yield_start)

crop_yields_50_years %>%
  filter(!is.na(code)) %>%
  ggplot(aes(yield_start, yield_end)) +
  geom_abline(color = "red") +
  geom_point() +
  facet_wrap(~ crop, scales = "free")

crop_yields_50_years %>%
  mutate(crop = fct_reorder(crop, yield_ratio)) %>%
  ggplot(aes(yield_ratio, crop)) +
  geom_boxplot() +
  scale_x_log10()

crop_yields_50_years %>%
  group_by(crop) %>%
  summarize(median_yield_ratio = median(yield_ratio)) %>%
  mutate(crop = fct_reorder(crop, median_yield_ratio)) %>%
  ggplot(aes(median_yield_ratio, crop)) +
  geom_col() +
  labs(title = "How much has the average country improved at producing this crop?",
       x = "(2018 yield) / (1968 yield)",
       y = "")

crop_yields_50_years %>%
  filter(is.na(code)) %>%
  filter(entity %in% c("Africa", "Asia", "Northern America", "South America", "Oceania")) %>%
  ggplot(aes(yield_start, yield_end, color = entity)) +
  geom_abline(color = "red") +
  geom_point() +
  expand_limits(y = 0, x = 0) +
  facet_wrap(~ crop, scales = "free") +
  labs(x = "Tonnes per hectare in 1968",
       y = "Tonnes per hectare in 2018",
       color = "Continent")
```

```{r}
library(ggrepel)
library(countrycode)

crop_yields_50_years %>%
  filter(crop == "Wheat",
         !is.na(code)) %>%
  mutate(continent = countrycode(code, "iso3c", "continent")) %>%
  filter(!is.na(continent)) %>%
  ggplot(aes(yield_start, yield_ratio)) +
  geom_point(aes(color = continent)) +
  scale_x_log10() +
  scale_y_log10(breaks = c(.25, .5, 1, 2, 4),
                labels = c("1/4X", "1/2X", "Same", "2X", "4X")) +
  geom_hline(yintercept = 1, linetype = "dotted") +
  geom_text_repel(aes(label = entity), force = .1,
                  size = 2.5) +
  labs(x = "1968 yield (tonnes per hectare), log scale",
       y = "(2018 yield) / (1968 yield), log scale",
       color = "Continent",
       title = "How has wheat efficiency changed across countries?")

countrycode("USA", "iso3c", "continent")
```