Replicating The Economist Plot
Tomy Tjandra
27 January 2020
1 Introduction
Hi! In this markdown, we’re going to replicate The Economist Plot from the daily chart entitled “How much would giving up meat help the environment?” using ggplot2 package. This plot give us an insight that going vegan for two-thirds of meals could cut food-related carbon emissions by 60%. We are going to break down the plot into two separate subplots and will be combined as one.
1.1 Import Packages
First, let’s us import some packages required for replicating the plot.
library(readxl) # read excel file
library(dplyr) # data wrangling
library(ggplot2) # data visualization
library(ggrepel) # for geom_text_repel
library(scales) # modify axes label
library(stringr) # string manipulation
library(png) # open png image
library(grid) # grid graphics for png
library(gridExtra) # additional function for grid package1.2 Load Font
Load all fonts from the device and use it to change the font family parameter of the plot.
library(extrafont)
#font_import()
loadfonts(device = "win")
custom_font_family <- "Segoe UI"2 Left Plot
This plot visualizes the health and environmental impact of one extra serving per day for each food type. The source of dataset is from PNAS research article “Multiple health and environmental impacts of foods”, by Clark et al., PNAS“.
2.1 Data Wrangling
Before we go any further with the visualization, let’s us prepare the data as follow:
Read the data from excel
Select columns of interest:
food_type,relative_risk, andenv_impactTransform
relative_riskfrom absolute into relative percentage
data_left <-
read_xlsx("datasets/pnas.1906908116.sd01.xlsx", sheet = 4, skip = 1) %>%
select(food_type = `Food Group`,
relative_risk = `Mean Relative Risk (RR) of All Cause Mortality`,
env_impact = `Average Relative Environmental Impact`) %>%
mutate(relative_risk = (relative_risk-1)*100)
head(data_left)## # A tibble: 6 x 3
## food_type relative_risk env_impact
## <chr> <dbl> <dbl>
## 1 "Whole \r\ngrains*" -8.00 1.62
## 2 "Nuts*" -24 4.03
## 3 "Legumes" -11.5 4.20
## 4 "Fruits*" -11.0 1.37
## 5 "Vegetables*" -13 1
## 6 "Refined \r\ngrains" -1. 1.622.1.1 String manipulation
We remove any whitespace and asterisk (*) character from the column food_type
# remove whitespace
data_left[1] <- lapply(data_left[1], function(x) gsub('\\s+', ' ', x))
# remove asterisk
data_left[1] <- lapply(data_left[1], function(x) gsub('\\*+', '', x))
head(data_left$food_type)## [1] "Whole grains" "Nuts" "Legumes" "Fruits"
## [5] "Vegetables" "Refined grains"2.1.2 Mapping food group
We need to create a mapping from food_type into food_group since the color of points corresponds to each individual food group, which are: Dairy, Eggs, Fish, Meat, Plant matter, and Sugars & Oils.
data_left_clean <-
data_left %>%
mutate(
food_group = as.factor(
case_when(
food_type %in% c("Whole grains", "Nuts", "Legumes", "Fruits", "Vegetables", "Refined grains", "Potatoes") ~ "Plant matter",
food_type %in% c("Chicken", "Unprocessed red meat", "Processed red meat") ~ "Meat",
food_type %in% c("SSBs", "Olive oil") ~ "Sugars & oils",
TRUE ~ as.character(food_type)
)
)
)
str(data_left_clean)## Classes 'tbl_df', 'tbl' and 'data.frame': 15 obs. of 4 variables:
## $ food_type : chr "Whole grains" "Nuts" "Legumes" "Fruits" ...
## $ relative_risk: num -8 -24 -11.5 -11 -13 ...
## $ env_impact : num 1.62 4.03 4.2 1.37 1 ...
## $ food_group : Factor w/ 6 levels "Dairy","Eggs",..: 5 5 5 5 5 5 5 3 1 2 ...2.2 Visualize
Now, it’s time for us to replicate the left plot using data_left_clean.
plot_left <-
ggplot(data = data_left_clean,
aes(x = relative_risk, y = env_impact)) +
geom_hline(yintercept = 0,
lwd = 0.75) +
geom_vline(xintercept = 0,
col = '#D85356',
lwd = 1) +
geom_point(aes(color = food_group), size = 5)
plot_left
Here, we create custom_color_palette to customize the color of geom_point() for each food_group.
custom_color_palette <- list(dairy = "#F4B265",
eggs = "#066DA0",
fish = "#2FBFD2",
meat = "#963C4C",
plant = "#70CAC6",
sugar = "#AC8D97")Then, change the scale of both x & y axes, add necessary labels, and use custom_color_palette which is already defined above.
plot_left <-
plot_left +
scale_x_continuous(limits = c(-31, 45),
expand = c(0, 0),
breaks = seq(-30, 40, by = 10)) +
scale_y_continuous(limits = c(0, 80),
expand = c(0, 0),
position = "right") +
labs(title = "Health and environmental impact\nof one extra serving per day",
x = "Relative risk of dying, %",
subtitle = "Average relative environmental impact*\n") +
scale_color_manual(values = unlist(custom_color_palette, use.names = FALSE)) +
coord_cartesian(clip = 'off')
plot_left
Apply theme to resemble the original plot.
plot_left <-
plot_left +
theme(
text = element_text(family = custom_font_family, size = 17,
color = "black"),
plot.title = element_text(face = "bold", size = 18),
plot.subtitle = element_text(hjust = 1),
axis.text.x = element_text(color = "black"),
axis.text.y = element_blank(),
axis.title.x = element_text(hjust = 0.43),
axis.title.y = element_blank(),
axis.ticks.length = unit(5, "pt"),
axis.ticks.x = element_line(size = 0.75),
axis.ticks.y = element_blank(),
panel.background = element_blank(),
panel.grid.minor = element_blank(),
panel.grid.major.x = element_blank(),
panel.grid.major.y = element_line(color = "#B2C2CA", size = 1),
legend.position = "none"
)
plot_left
Manually add the y-axis inside the plot by using geom_text().
plot_left <-
plot_left +
geom_text(
data = data.frame(env_impact = seq(0, 80, 20)),
aes(label = env_impact),
x = 45,
hjust = 1,
vjust = -0.4
)
plot_left
2.2.1 Highlight
As we can see from the plot, there are exactly 10 points which are being highlighted. Here, we create highlight_color to customize the color of line for each food_group.
highlight_color <- list(dairy = "#EAB651",
eggs = "#006399",
fish = "#78ADCA",
meat = "#BE838A",
plant = "#3EADB0")Create a new data frame called data_left_highlight to store the label together with its nudge_x and nudge_y properties.
data_left_highlight <- data.frame(
food_type = c("Chicken", "Potatoes", "Fish",
"Dairy", "Refined grains",
"Vegetables",
"Nuts", "Eggs",
"Unprocessed red meat", "Processed red meat"),
nudge_x = c(0, -3, 0, 6.5, 9, -7,
-4, 2, 2, -1),
nudge_y = c(35, 35, 20, 10, 10, 17,
4, 1, -3, -5)
)We loop through each row of data_left_highlight to highlight the points one by one.
plot_left_highlight <- plot_left
for (row in 1:nrow(data_left_highlight)) {
data_highlight <-
data_left_clean %>%
filter(food_type == data_left_highlight$food_type[row])
if (row <= 6) {
# text with segment line (repel)
plot_left_highlight <-
plot_left_highlight +
geom_text_repel(
aes(label = food_type,
family = custom_font_family),
size = 5,
data = data_highlight,
segment.color = highlight_color[data_highlight$food_group],
segment.size = 1,
nudge_x = data_left_highlight$nudge_x[row],
nudge_y = data_left_highlight$nudge_y[row],
direction = "y"
)
} else {
# text without segment line
plot_left_highlight <-
plot_left_highlight +
geom_text(
aes(label = str_wrap(food_type, 10),
family = custom_font_family),
size = 5,
data = data_highlight,
nudge_x = data_left_highlight$nudge_x[row],
nudge_y = data_left_highlight$nudge_y[row],
hjust = ifelse(row == 10, 1, 0)
)
}
plot_left_highlight <-
plot_left_highlight +
geom_point(
data = data_highlight,
shape = 21, size = 5,
color = highlight_color[data_highlight$food_group],
fill = custom_color_palette[data_highlight$food_group])
}
plot_left_highlight
2.3 Result
Finally, we add black line as a header on top-left of the plot and export it as plot_left.png.
png("assets/plot_left.png", width = 7, height = 6, units = "in", res = 300)
plot_left_highlight
# Add header: black line
grid.rect(x = 0.0575, y = 0.995,
hjust = 1, vjust = 0,
width = 0.05,
gp = gpar(fill="#353535",lwd=0))
dev.off()## png
## 2
3 Right plot
This plot visualizes the greenhouse-gas footprint (kg of CO2 equivalent per person per year) in United States for each diet type and food group. The source of dataset is from ScienceDirect article “Country-specific dietary shifts to mitigate climate and water crises, by Kim et al., Global Environmental Change”.
3.1 Data Wrangling
Before we go any further with the visualization, let’s us prepare the data:
Read the data from csv file
Filter the data which country is “United States of America” and attribute is the greenhouse-gas footprint which stored as “kg_co2e_total”
Filter
diet_typebased on plot: “Average diet”, “Reduced diet”, “No dairy”, “Vegetarian”, “Two-thirds vegan”, and “Vegan”Select columns of interest:
diet,food_type, andvalue
diet_type <- c("baseline", "baseline_adjusted", "no_dairy", "lacto_ovo_vegetarian", "2/3_vegan", "vegan")
data_right_clean <-
read.csv("datasets/diet_footprints_by_country_diet_output_group.csv") %>%
filter(country == "United States of America",
attribute == "kg_co2e_total",
diet %in% diet_type) %>%
select(diet, food_type = output_group, value)
head(data_right_clean)## diet food_type value
## 1 2/3_vegan Aquatic animals 20.92218
## 2 2/3_vegan Bovine meat 240.83332
## 3 2/3_vegan Dairy 138.51703
## 4 2/3_vegan Eggs 20.42052
## 5 2/3_vegan Fruits 22.46829
## 6 2/3_vegan Grains 62.136813.1.1 Mapping food group
Just like the left plot, we need to create a mapping from food_type into food_group since the color of bars corresponds to each individual food group, which are: Dairy, Eggs, Fish, Meat, Plant matter, and Sugars & Oils. Then we drop food_type since it’s not being used for our plot.
data_right_clean <-
data_right_clean %>%
mutate(
food_group = as.factor(
case_when(
food_type %in% c("Aquatic animals") ~ "Fish",
food_type %in% c("Bovine meat", "Insects", "Offals", "Pig meat", "Poultry", "Sheep and goat meat") ~ "Meat",
food_type %in% c("Fruits", "Grains", "Nuts and seeds", "Pulses and soy", "Starchy roots", "Vegetables") ~ "Plant matter",
food_type %in% c("Sugars", "Vegetable oils") ~ "Sugars & oils",
TRUE ~ as.character(food_type)
)
)
) %>%
select(-food_type)
str(data_right_clean)## 'data.frame': 89 obs. of 3 variables:
## $ diet : Factor w/ 12 levels "2/3_vegan","baseline",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ value : num 20.9 240.8 138.5 20.4 22.5 ...
## $ food_group: Factor w/ 6 levels "Dairy","Eggs",..: 3 4 1 2 5 5 4 5 4 4 ...3.2 Visualize
Now, it’s time for us to replicate the right plot using data_right_clean.
plot_right <-
ggplot(data = data_right_clean,
aes(x = reorder(diet, value, sum), y = value)) +
geom_col(aes(fill = food_group),
position = position_stack(reverse = TRUE),
width = 0.75) +
geom_hline(yintercept = 0,
lwd = 1.5) +
coord_flip()
plot_right
Here, we change the scale of both x-y axes, add necessary labels, and use custom_color_palette which is already defined above.
plot_right <-
plot_right +
scale_x_discrete(labels = str_wrap(rev(c("Average diet", "Reduced diet*", "No diary*", "Vegetarian*", "Two-thirds vegan*", "Vegan*")),
width = 15)) +
scale_y_continuous(limits = c(0, 2600),
expand = c(0, 0),
position = "right",
labels = comma) +
labs(title = "United States, greenhouse-gas footprint",
subtitle = expression(paste("kg of ", CO[2], " equivalent per person per year"))) +
scale_fill_manual(values = unlist(custom_color_palette, use.names = FALSE))
plot_right
Apply theme to resemble the original plot.
plot_right <-
plot_right +
theme(
text = element_text(family = custom_font_family, size = 17),
plot.title = element_text(face = "bold", size = 18),
plot.title.position = "plot",
plot.subtitle = element_text(face = "bold"),
axis.text = element_text(color = "black"),
axis.text.y = element_text(hjust = 0),
axis.title = element_blank(),
axis.ticks = element_blank(),
panel.background = element_blank(),
panel.grid.minor = element_blank(),
panel.grid.major.x = element_line(color = "#B2C2CA", size = 1),
panel.grid.major.y = element_blank(),
legend.title = element_blank(),
legend.position = "top",
legend.direction = "horizontal"
) +
guides(fill = guide_legend(nrow = 1))
plot_right
3.3 Result
Finally, we shift the legend to the left, add black line as a header on top-left of the plot, and export it as plot_right.png.
png("assets/plot_right.png", width = 7, height = 6, units = "in", res = 300)
# Shift legend
plot_right <-
plot_right +
theme(
legend.spacing.x = unit(3, units = "pt"),
legend.text = element_text(margin = margin(
r = 15, unit = "pt")),
legend.justification = c(0, 0.875)
)
gt <- ggplot_gtable(ggplot_build(plot_right))
gb <- which(gt$layout$name == "guide-box")
gt$layout[gb, 1:4] <- c(1, 1, max(gt$layout$b), max(gt$layout$r))
grid.newpage()
grid.draw(gt)
# Add header: black line
grid.rect(x = 0.0575, y = 0.995,
hjust = 1, vjust = 0,
width = 0.05,
gp = gpar(fill="#353535",lwd=0))
dev.off()## png
## 2
4 Final Result
Last but foremost, we combine both of the plot as one png file. The steps are as follow:
Read
plot_left.pngandplot_right.pngArrange the two plot as one
pngAdd red line as header, which is the signature of The Economist Plot
Add title and caption as needed
png("assets/plot_combined.png", width = 13, height = 7, units = "in", res = 300)
img1 <- rasterGrob(as.raster(readPNG("assets/plot_left.png")), interpolate = FALSE)
img2 <- rasterGrob(as.raster(readPNG("assets/plot_right.png")), interpolate = FALSE)
spacing <- rectGrob(gp=gpar(col="white"))
# Plot
grid.arrange(img1, spacing, img2, ncol = 3,
widths = c(0.49, 0.02, 0.49))
# Economist red line
grid.rect(x = 1, y = 0.995,
hjust = 1, vjust = 0,
gp = gpar(fill='#E5001C',lwd=0))
grid.rect(x = 0.04, y = 0.98,
hjust = 1, vjust = 0,
gp = gpar(fill='#E5001C',lwd=0))
# Title
grid.text("High-steak diets",
x = 0.165, y = 0.94,
hjust = 1, vjust = 0,
gp = gpar(fontsize=20, fontfamily=custom_font_family,
fontface="bold"))
# Caption (left)
caption_left <- 'Sources: "Multiple health and environmental impacts of foods", by Clark et al., PNAS; "Country-specific\ndietary shifts to mitigate climate and water crises", by Kim et al., Global Environmental Change'
grid.text(caption_left,
x = 0, y = 0.02,
hjust = 0, vjust = 0,
gp = gpar(fontsize=15, fontfamily=custom_font_family,
col="#5E5E5E"))
# Caption (right, top)
caption_right_top <- expression(paste(""^"*", "Vegetables=1\t",
""^"+", "Simulated diet, to"))
grid.text(caption_right_top,
x = 0.995, y = 0.055,
hjust = 1, vjust = 0,
gp = gpar(fontsize=15, fontfamily=custom_font_family,
col="#5E5E5E"))
# Caption (right, bottom)
grid.text("reach 2,300 calories per day",
x = 0.995, y = 0.02,
hjust = 1, vjust = 0,
gp = gpar(fontsize=15, fontfamily=custom_font_family,
col="#5E5E5E"))
dev.off()## png
## 2Voilà, here it is! We successfully replicate The Economist Plot by using ggplot2 package.
