Data Sources

The data used in this workshop are adapted from the following:

City of Cambridge Public Safety. “Police Department Crash Data - Updated.” Cambridge Open Data, September 13, 2022. https://data.cambridgema.gov/Public-Safety/Police-Department-Crash-Data-Updated/gb5w-yva3.

National Centers for Environmental Information, National Oceanic and Atmospheric Administration. “Past Weather.” Past Weather | CAMBRIDGE MA | US1MAMD0011, October 3, 2022. https://www.ncei.noaa.gov/access/past-weather/42.55502054693938,-71.76079845408827,42.040922629398835,-70.48652687279844.

Plotting with ggplot2

library(tidyverse)
library(lubridate)

Practice with irises

ggplot(iris, aes(x = Petal.Length, y = Petal.Width)) +
  geom_point()

Start with just the data and coordinates

ggplot(iris, aes(x = Petal.Length, y = Petal.Width))

Add in points to show data, along with color and a linear model

ggplot(iris, aes(x = Petal.Length, y = Petal.Width)) +
  geom_point()

ggplot(iris, aes(x = Petal.Length, y = Petal.Width, color = Species)) +
  geom_point() 

ggplot(iris, aes(x = Petal.Length, y = Petal.Width, color = Species)) +
  geom_point() +
  geom_smooth(method = "lm")

ggplot(iris, aes(x = Petal.Length, y = Petal.Width)) +
  geom_point(aes(color = Species)) +
  geom_smooth(method = "lm")

Activity 1

Working with real data

Read in our data

crashes <- read_csv("./data/processed/crashes.csv")
weather <- read_csv("./data/processed/weather.csv")

crashes_weather <- crashes %>%
  left_join(weather)

Let’s interrogate whether those match up sensibly

ggplot(crashes_weather, aes(x = weather_condition, y = precip)) + 
  geom_boxplot()

Good enough for practice! We’d want to be more careful if this were an actual analysis.

Start simple, single variable

Histograms

ggplot(crashes_weather, aes(x = precip)) +
  geom_histogram()

Adjust bins/binwidth as necessary

ggplot(crashes_weather, aes(x = precip)) +
  geom_histogram(bins = 40)

ggplot(crashes_weather, aes(x = precip)) +
  geom_histogram(bins = 10)

ggplot(crashes_weather, aes(x = precip)) +
  geom_histogram(bins = 20)

Quick color adjustment

Color is different from fill

ggplot(crashes_weather, aes(x = precip)) +
  geom_histogram(bins = 20, color = "black", fill = "steelblue")

Activity 2

How does this compare to the overall histogram for our weather data set?

ggplot(weather, aes(x = precip)) +
  geom_histogram(bins = 20, color = "black", fill = "steelblue")

Categorical variables

That actually matches pretty closely - there are just more days with less precipitation. Let’s try looking at number of records based on reported weather condition.

Bar plots, defaults to stat = “count”

ggplot(crashes_weather, aes(x = weather_condition)) +
  geom_bar()
ggplot(crashes_weather, aes(x = weather_condition)) +
  geom_bar() +
  theme(
    axis.text.x = element_text(angle = 45)
  )

dplyr to calculate crashes/day based on if it rained

Let’s try something a little advanced. We’ll make a data set to see whether there was precipitation on a given day, and then compare crashes per day for days with precipitation and without precipitation.

crashes_per_day <- crashes_weather %>%
  group_by(date, precip) %>%
  summarize(crashes = n()) %>%
  mutate(
    was_precip = precip > 0,
    year = year(date)
  )

knitr::kable(head(crashes_per_day))

Boxplots

ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_boxplot()

Violin Plots

ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin()

ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin() +
  geom_jitter(alpha = 0.2)

Other features of ggplot2

Starting with basic scatterplot

ggplot(crashes_per_day, aes(x = precip, y = crashes)) +
  geom_point()

Plotting in Layers

Points created first, line drawn on top

ggplot(crashes_per_day, aes(x = precip, y = crashes)) +
  geom_point() +
  geom_smooth()

Line created first, points drawn on top

ggplot(crashes_per_day, aes(x = precip, y = crashes)) +
  geom_smooth() +
  geom_point()

Using colors

ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin() +
  geom_jitter(alpha = 0.2)

This applies the color to everything - global ggplot call carries to all geom_*

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = was_precip)) + 
  geom_violin() +
  geom_jitter(alpha = 0.2)

If we want to color the violins based on was_precip:

  • This doesn’t work
ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin(color = was_precip) +
  geom_jitter(alpha = 0.2)
  • This does! Using the data to color the geom_* requires the aes() function
ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin(aes(color = was_precip)) +
  geom_jitter(alpha = 0.2)

This changes to a color not defined by the data (also see alpha in the geom_jitter)

ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin(color = "orange") +
  geom_jitter(alpha = 0.2)

This creates a color based on one value, “orange.” It doesn’t involve the color orange at all.

ggplot(crashes_per_day, aes(x = was_precip, y = crashes)) + 
  geom_violin(aes(color = "orange")) +
  geom_jitter(alpha = 0.2)

Colors can also be continuous rather than categorical

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = crashes)) + 
  geom_violin() +
  geom_jitter(alpha = 0.2)

Labels, Legends, and Themes

Titles and Labels

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = precip)) + 
  geom_violin() +
  geom_jitter(alpha = 0.5) +
  labs(
    x = "Was there Precipitation?",
    y = "Number of Crashes",
    title = "Vehicle Crashes and Precipitation",
    subtitle = "In Cambridge, MA",
    caption = "Data from NOAA and City of Cambridge",
    color = "Amount of Precipitation"
  )

Legend / Color Customization

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = precip)) + 
  geom_violin() +
  geom_jitter(alpha = 0.5) +
  labs(
    x = "Was there Precipitation?",
    y = "Number of Crashes",
    title = "Vehicle Crashes and Precipitation",
    subtitle = "In Cambridge, MA",
    caption = "Data from NOAA and City of Cambridge",
    color = "Amount of Precipitation"
  ) +
  scale_color_continuous(
    low = "yellow", 
    high = "blue"
  )

Using built-in themes

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = precip)) + 
  geom_violin() +
  geom_jitter(alpha = 0.5) +
  labs(
    x = "Was there Precipitation?",
    y = "Number of Crashes",
    title = "Vehicle Crashes and Precipitation",
    subtitle = "In Cambridge, MA",
    caption = "Data from NOAA and City of Cambridge",
    color = "Amount of Precipitation"
  ) +
  scale_color_continuous(
    low = "yellow", 
    high = "blue"
  ) +
  theme_dark()

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = precip)) + 
  geom_violin() +
  geom_jitter(alpha = 0.5) +
  labs(
    x = "Was there Precipitation?",
    y = "Number of Crashes",
    title = "Vehicle Crashes and Precipitation",
    subtitle = "In Cambridge, MA",
    caption = "Data from NOAA and City of Cambridge",
    color = "Amount of Precipitation"
  ) +
  scale_color_continuous(
    low = "yellow", 
    high = "blue"
  ) +
  theme_bw()

Facets / Small Multiples

ggplot(crashes_per_day, aes(x = was_precip, y = crashes, color = precip)) + 
  geom_violin() +
  geom_jitter(alpha = 0.5) +
  labs(
    x = "Was there Precipitation?",
    y = "Number of Crashes",
    title = "Vehicle Crashes and Precipitation",
    subtitle = "In Cambridge, MA",
    caption = "Data from NOAA and City of Cambridge",
    color = "Amount of Precipitation"
  ) +
  scale_color_continuous(
    low = "blue", 
    high = "red"
  ) +
  theme_bw() +
  facet_wrap( ~ year)