Lecture 2.3 KEY

Visualizing evolution

Author

Your Name

Published

January 13, 2025

Note

This template follows lecture 2.3 slides. Please be sure to cross-reference the slides, which contain important information and additional context!

Setup

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                                    setup                                 ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

#.........................load libraries.........................
library(tidycensus)
library(tidyverse)
library(gghighlight)

#..........................import data...........................
lyme <- read_csv(here::here("week2", "data", "Lyme_Disease_Cases_by_State_or_Locality.csv"))

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                          wrangle lyme disease data                       ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

#....................wrangle lyme disease data...................
lyme_clean <- lyme |> 
  
  rename(state = State) |> 
  
  # remove special character from State names that have it ----
  mutate(state = str_replace(string = state, pattern = "\x86", replacement = "")) |> 
  
  # remove total cases row ----
  filter(state != "U.S. Total") |> 
  
  # wide to long (tidy) years ----
  pivot_longer(cols = 2:16, names_to = "year", values_to = "reported_cases") |> 
  
  # reorder columns ----
  relocate(year, state, reported_cases) |> 
  
  # filter for years 2010 - 2020 ----
  filter(year %in% c(2010:2020))
  
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                          get & wrangle census data                       ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

#........get pop estimates by state using tidycensus API.........
us_state_pop <- get_estimates(geography = "state", 
                              product = "population",
                              state = NULL, 
                              year = 2019) |> # request 2019 vintage for the 2010s (see documentation)
  filter(variable == "POP") |> 
  select(state = NAME, population = value) |> 
  filter(state != "Puerto Rico")

#........................write data to csv.......................
# OPTIONAL, but recommended in case you want to work offline, the API is down, etc. (you can then read in your saved data file rather than run the above code)
# write_csv(us_state_pop, file = here::here("week2", "data", "us_state_pop.csv"))  

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                            join lyme & pop dfs                           ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

lyme_pop <- left_join(lyme_clean, us_state_pop) |>  
  
  # add col with num of 100k people per state ----
  mutate(pop100k = population/100000) |> 
  
  # calculate num cases per 100k people (common way of reporting disease incidence) ----
  mutate(cases_per100k = reported_cases/pop100k)

Line graphs

  • show evolution of one or more numeric variables; similar to scatter plots but points are ordered (typically by x-axis value) and joined with straight line segments
# line chart ----
ggplot(lyme_pop, aes(x = year, y = cases_per100k, group = state)) + 
  geom_line() +
  geom_point()

Alt 1: Avoid spaghetti plots

# highlight one group ----
ggplot(lyme_pop, aes(x = year, y = cases_per100k, group = state)) +
  geom_line() +
  gghighlight::gghighlight(state == "New Jersey") 

# highlight based on criteria ----
ggplot(lyme_pop, aes(x = year, y = cases_per100k, group = state, color = state)) +
  geom_line() +
  gghighlight::gghighlight(max(cases_per100k) > 100)

Note 1: It’s okay to cut the y-axis

# ggplot cuts y-axis by default ----
lyme_pop |> 
  filter(year %in% c(2013:2019)) |> 
  filter(state == "Vermont") |> 
  ggplot(aes(x = year, y = cases_per100k, group = state)) +
  geom_line()
# force y-axis origin at 0 ----
lyme_pop |> 
  filter(year %in% c(2013:2019)) |> 
  filter(state == "Vermont") |> 
  ggplot(aes(x = year, y = cases_per100k, group = state)) +
  geom_line() +
  scale_y_continuous(limits = c(0, NA))

Note 2: Consider aspect ratio

  • here, 10 units on our x-axis (years 2010 - 2020), and ~175 units (case_per100k 0 to ~175) on our y-axis
  • ggplot automatically adjusts aspect ratio so that plot is easier to read
# default aspect ratio (with added tick marks) ----
lyme_pop |> 
  filter(state == "Vermont") |> 
  ggplot(aes(x = year, y = cases_per100k, group = state)) +
  geom_line() +
  scale_y_continuous(breaks = seq(0, 190, by = 1))

  • we can fix the aspect ratio of our plot
  • ratio is expressed as y / x and is set to 1 by default (see grid lines)
# fixed aspect ratio ----
lyme_pop |> 
  filter(state == "Vermont") |> 
  ggplot(aes(x = year, y = cases_per100k, group = state)) +
  geom_line() +
  scale_y_continuous(breaks = seq(0, 190, by = 1)) +
  coord_fixed(ratio = 1)

  • ratios > 1: units on the y-axis longer than units on the x-axis (resulting in steeper slopes)
  • ratios < 1: units on the x-axis longer than units on the y-axis (resulting in shallower slopes)
# adjusted aspect ratio ----
lyme_pop |> 
  filter(state == "Vermont") |> 
  ggplot(aes(x = year, y = cases_per100k, group = state)) +
  geom_line() + 
  coord_fixed(ratio = 1/50)

Area charts

  • similar to line chart; indicates change in a numeric variable through time, but the space between x-axis and line is shaded in
# area chart of lyme disease incidence in NJ ----
lyme_pop |> 
  filter(state == "New Jersey") |> 
  ggplot(aes(x = year, y = cases_per100k, group = state, fill = state)) +
  geom_area() +
  scale_fill_manual(values = c("#047C90")) +
  scale_x_discrete(expand = c(0, 0)) +
  scale_y_continuous(expand = c(0, 0)) +
  theme(legend.position = "none")

Stacked area charts

  • show evolution of the whole + relative proportions of each group
# stacked area chart ----
lyme_pop |> 
  filter(state %in% c("Maine", "Rhode Island", "New Hampshire", "Vermont")) |> 
  ggplot(aes(x = year, y = cases_per100k, group = state, fill = state)) +
  geom_area()

Alt 1: proportional stacked area chart

  • plot percentage contributions rather than absolute numbers on y-axis
# proportional stacked area chart ----
lyme_pop |> 
  filter(state %in% c("Maine", "Rhode Island", "New Hampshire", "Vermont")) |> 
  ggplot(aes(x = year, y = cases_per100k, group = state, fill = state)) +
  geom_area(position = "fill") +
  scale_y_continuous(labels = scales::label_percent(scale = 100))

Note 1: stacked area charts aren’t great for looking at evolution of individual groups

# stacked area chart ----
lyme_pop |> 
  filter(state %in% c("Maine", "Rhode Island", "New Hampshire", "Vermont")) |> 
  ggplot(aes(x = year, y = cases_per100k, group = state, fill = state)) +
  geom_area()
# same data, but as a line chart & faceted by state ----
lyme_pop |> 
  filter(state %in% c("Maine", "Rhode Island", "New Hampshire", "Vermont")) |> 
  ggplot(aes(x = year, y = cases_per100k, group = state, color = state)) +
  geom_line() +
  facet_wrap(~state)