Lecture 3: Exercise Problems and Solutions

Load packages

library(ggplot2)
library(data.table)
library(dplyr)

1 Exercise 1

Let’s use the economics data, which is a dataset built into the ggplot2 package. It was produced from US economic time series data available from Federal Reserve Economic Data. This contains the following variables:

  • date: date in year-month format
  • pce: personal consumption expenditures, in billions of dollars
  • pop: total population in thousands
  • psavert: personal savings rate
  • uempmed: median duration of unemployment in weeks
  • unemploy: number of unemployed in thousands

1. Create a scatter plot of unemploy (x-axis) and psavert (y-axis). Add a simple regression line to the plot. Change the x-axis, y-axis, and fill legend labels to something more informative.

2. Create a bar plot of psavert by date. Use pop for fill color. Change the x-axis, y-axis, and fill legend labels to something more informative.

  • Hint: use stat = 'identity' in the geom_bar() function to plot the actual values of pce.

3. (Challenging) Create a multiple line plot taking day as x-axis and psavert and uempmed as y-axis, respectively. The output should look like the following.

  • Hint: I think there are multiple ways to do this.
# === Part 1 === #
ggplot(data = economics, aes(x = unemploy, y = psavert)) +
  geom_point()+
  geom_smooth(method = "lm", formula = y ~ x) +
  labs(
    x = "Unemployment (thousands)",
    y = "Personal savings rate",
    title = "Relationship between unemployment and personal savings rate"
  )

# === Part 2 === #
ggplot(data = economics) +
  geom_bar(aes(x = date, y = psavert, fill = pop), stat = 'identity') +
  labs(
    x = "Date",
    y = "Personal Savings Rate",
    fill = 'Population',
    title = "Personal Consumption Expenditures by Date"
  )

# === Part 3 === #
ggplot(data = economics) +
  geom_line(aes(x = date, y = psavert, color = "Personal Savings Rate")) +
  geom_line(aes(x = date, y = uempmed, color = "Median Duration of Unemployment"))

# In the code above, I used two geom_line() functions to create two lines in the same plot. Although there is no color called "Personal Savings Rate", if you use `color=<color_name>` option inside the `aes()` function, ggplot regards <color_name> as a variable and assigns color automatically. Also, it will be shown in the legend.


# Alternative solution (This is more efficient if you have many variables to plot).
economics_long <-
  as.data.table(economics) %>%
  melt(
    id.vars = "date",
    measure.vars = c("psavert", "uempmed")
  ) %>%
  .[, new_names := ifelse(variable == "psavert", "Personal Savings Rate", "Median Duration of Unemployment")]

ggplot(economics_long) +
  geom_line(
    aes(x = date, y = value, color = new_names)
  )+
  labs(
    x = "Date",
    y = "Value",
    color = "Variable"
  )

2 Exercise 2

For this exercise problem, we will use medical cost personal datasets descried in the book “Machine Learning with R” by Brett Lantz. The dataset provides \(1,338\) records of medical information and costs billed by health insurance companies in 2013, compiled by the United States Census Bureau.

The dataset contains the following variables:

  • age: age of primary beneficiary
  • sex: insurance contractor gender, female, male
  • bmi: body mass index, providing an understanding of body, weights that are relatively high or low relative to height
  • children: number of children covered by health insurance
  • smoker: smoking
  • region: the beneficiary’s residential area in the US; northeast, southeast, southwest, northwest.
  • charges: individual medical costs billed by health insurance

Download the data

# === Download Data (Don't worry about this part.) === #
insurance_url <- "https://raw.githubusercontent.com/stedy/Machine-Learning-with-R-datasets/master/insurance.csv"
insurance <- rio::import(insurance_url)

# === Take a look at the data === #
head(insurance)
  age    sex    bmi children smoker    region   charges
1  19 female 27.900        0    yes southwest 16884.924
2  18   male 33.770        1     no southeast  1725.552
3  28   male 33.000        3     no southeast  4449.462
4  33   male 22.705        0     no northwest 21984.471
5  32   male 28.880        0     no northwest  3866.855
6  31 female 25.740        0     no southeast  3756.622

  1. Create a histogram of charges by sex in the same plot. Fill the boxes with different colors for each sex.

  2. Create a scatter plot of bmi (x-axis) and charges (y-axis).

  3. Now, create a scatter plot of bmi (x-axis) and charges (y-axis), and add regression lines by smoke (So, there are two regression lines: one for group of smokers and the other for group of non-smokers).

  4. Create the following plot.

# === Part 1 === #
ggplot(data = insurance) +
  geom_histogram(aes(x = charges, fill = sex), alpha = 0.5) +
  labs(
    x = "Medical Charges",
    y = "Frequency",
    title = "Histogram of individual medical costs by sex"
  )


# === Part 2 === #
ggplot(data = insurance) +
  geom_point(aes(x = bmi, y = charges)) +
  labs(
    x = "Body Mass Index (BMI)",
    y = "Scatter plot of BMI vs medical costs"
  )

# === Part 3 === #
ggplot(data = insurance, aes(x = bmi, y = charges, color = smoker)) +
  geom_point()+
  geom_smooth(method = "lm", formula = y ~ x) +
  labs(
    x = "Body Mass Index (BMI)",
    y = "Scatter plot of BMI vs medical costs by smoker Status",
    color = "Smoker"
  )

# === Part 4 === #
ggplot(data = insurance) +
  geom_boxplot(aes(x = sex, y = charges, fill = region)) +
  labs(
    x = "Sex",
    y = "Medical costs",
    title = "Distribution of individual medical mosts by sex and region"
  )

3 Exercise 3

For this exercise problem, you will use the gapminder data from the gapminder package.

# install the gapminder package if you haven't done so.
# install.packages("gapminder")

# Load the data from the gapminder package
library(gapminder)
data(gapminder)

gapminder <- as.data.table(gapminder) # Convert the data to a data.table object (only if you want)
# or
# setDT(gapminder)

# See ?gapminder for more information about the dataset.

  1. Find the number of unique countries in the data.

  2. Calculate the mean life expectancy for the entire dataset.

  3. Create a dataset by subsetting the data for the year 2007. Create a scatter plot of GDP per capita vs. life expectancy for the year 2007, color-coded by continent.

  4. Create a bar plot showing the total population for each continent in 2007. Fill the bars with blue and set the transparency to 0.5.

  5. Subset the data for the United States, China, India, and the United Kingdom. Create a line plot showing the change in life expectancy over time for these countries.

  6. Create a scatter plot of GDP per capita vs. life expectancy for the entire gapminder dataset. Use facet_wrap to create separate plots for each continent.

  7. Group the data by continent and calculate the mean GDP per capita for each continent for each year. Create a line plot showing the trend of mean GDP per capita for each continent over time.

# === Part 1 === #
#' There are several approaches.
#' The most straightforward way is to use the unique() function, and then count the number of unique values.
length(unique(gapminder$country))

# Alternatively, you could use the uniqueN() function from the data.table package.
# gapminder[, uniqueN(country)]

# === Part 2 === #
mean(gapminder$lifeExp)
# same as
# gapminder[, mean(lifeExp)]

# === Part 3 === #
gapminder_2007 <- gapminder[year == 2007,]

ggplot(data = gapminder_2007) +
  geom_point(aes(x = gdpPercap, y = lifeExp, color = continent)) +
  labs(
    x = "GDP per capita",
    y = "Life expectancy",
    title = "Scatter plot of GDP per capita vs. life expectancy in 2007"
  )

# === Part 4 === #
ggplot(data = gapminder_2007) +
  geom_bar(aes(x = continent, y = pop), stat = "identity", fill = "blue", alpha = 0.5) +
  labs(
    x = "Continent",
    y = "Total population",
    title = "Total population by continent in 2007"
  )


# === Part 5 === #
gapminder[country %in% c("United States", "China", "India", "United Kingdom"),] %>%
  ggplot(data = .)+
  geom_line(aes(x = year, y = lifeExp, color = country))+
  labs(
    title = "Life Expectancy in the selected countries",
    x = "Year", y = "Life Expectancy"
  )+
  theme_bw()+
  theme(plot.title = element_text(hjust = 0.5))

# === Part 6 === #
ggplot(data = gapminder) +
  geom_point(aes(x = gdpPercap, y = lifeExp, color = continent)) +
  facet_wrap(vars(continent), scales = "fixed") +
  labs(
    x = "GDP per capita",
    y = "Life expectancy",
    title = "Scatter plot of GDP per capita vs. life expectancy by continent"
  )

# === Part 7 === #
gapminder[, .(mean_gdp = mean(gdpPercap)), by = .(continent, year)] %>%
  ggplot(data = .) +
  geom_line(aes(x = year, y = mean_gdp, color = continent)) +
  labs(
    x = "Year",
    y = "Mean GDP per capita",
    title = "Trend of mean GDP per capita by continent"
  )

4 Exercise 4

For this exercise problem, we will use economics dataset from the ggplot2 package. You need to use data manipulation and visualization techniques using the data.table and ggplot2 packages.

# === Loading data === #
economics <- as.data.table(ggplot2::economics)
  1. As you already know by now, the economics dataset contains various economic indicators for the United States. We want to create a line plot showing the trends of all economic indicators over time. Each economic indicator is stored in a separate column in the data, and you can visualize each indicator by creating a single line plot, separately. But, there is a better way to do this. It should look like the following plot.

# Since we want use column names as variable names, we need to convert the data to a long format. We melt all the columns except the date column.
economics_long <- melt(economics, id.vars = "date")

# Then, create a facet plot where each panel shows the trend of one economic indicator.
# Since we only have one variable to facet, facet_wrap() function is more appropriate.
# Also, we set scales = "free_y" to make the y-axis scales free across facets.
ggplot(data = economics_long) +
  geom_line(aes(x = date, y = value)) +
  facet_wrap(vars(variable), scales = "free_y")

5 Exercise 5 (challenging)

For this exercise problem, you will use “corn_yield_dt.rds” in the “Data” folder. I obtained this from USDA-NASS Quick Stats database. The data contains the county-level corn yield data (in BU / ACRE) for each major corn production state in the US Midwest from 2000 to 2022.

  1. Load the data and take a look at it.

  2. Convert the data to a data.table object. The Value column contains the corn yield data. Rename the column to yield.

  3. Let’s derive the state-level annual average corn yield data by calculating the mean of corn yield by state and year. Create a line plot of the annual trend of corn yield in Minnesota by taking year for the x-axis and the derived mean yield for the y-axis.

  4. Create line plots showing the trend of annual corn yield for each state in the same plot.

  5. Create a facet plot showing each state’s annual corn yield trend. To compare the trends across states, use scales = "fixed".

Hint: state_alpha is the two-letter state abbreviation for each state.

  1. Create a new dataset that contains the overall average corn yield across states by taking the mean of the yield by year. Add a line plot of this dataset to the plot you created in the previous step. Use red dashed line to represent this line.
  • If you could add a legend to the plot to indicate what the red dashed line means, that would be great! To do this, you need to use scale_color_manual() function.
# === Part 1 === #
yield_dt <- readRDS("Data/corn_yield_dt.rds")

# === Part 2 === #
yield_dt <- 
  as.data.table(yield_dt) %>% 
  setnames("Value", "yield")

# === Part 3 === #
mean_yield_state_y_dt <- 
  yield_dt %>%
  .[, .(mean_yield = mean(yield, na.rm = TRUE)), by = .(state_alpha, year)]

mean_yield_state_y_dt %>%
  .[state_alpha == "MN",] %>%
  ggplot(data = .) +
    geom_line(aes(x = year, y = mean_yield)) +
    labs(
      x = "Year",
      y = "Mean corn yield (BU / ACRE)",
      title = "Average corn yield produced in Minnesota"
    )


# === Part 4 === #
ggplot(data = mean_yield_state_y_dt) +
  geom_line(aes(x = year, y = mean_yield, color = state_alpha)) +
  labs(
      x = "Year",
      y = "Mean corn yield (BU / ACRE)",
      title = " Corn yield trend in Minnesota"
    )

# === Part 5 === #
ggplot(data = mean_yield_state_y_dt) +
  geom_line(aes(x = year, y = mean_yield)) +
  facet_wrap(vars(state_alpha), scales = "fixed") +
  labs(
      x = "Year",
      y = "Mean corn yield (BU / ACRE)",
      title = "Corn yield trend by state"
    )

# === Part 6 === #
mean_yield_year_dt <- 
  yield_dt %>%
  .[, .(mean_yield = mean(yield)), by = .(year)]


ggplot() +
  geom_line(data = mean_yield_state_y_dt, aes(x = year, y = mean_yield)) +
  facet_wrap(vars(state_alpha), scales = "fixed") +
  geom_line(data = mean_yield_year_dt, aes(x = year, y = mean_yield), linetype = "dashed", color = "red") +
  labs(
      x = "Year",
      y = "Corn yield (BU / ACRE)",
      title = "Corn yield trend by state"
    ) +
  scale_color_manual(
    values = c("Average corn yield across states" = "red")
    ) +
  theme_bw() +
  theme(
    legend.position = "bottom"
  )


# Furthermore, you can add a legend to the plot by using the following code.
# ggplot() +
#   geom_line(data = mean_yield_state_y_dt, aes(x = year, y = mean_yield)) +
#   facet_wrap(vars(state_alpha), scales = "fixed") +
#   geom_line(data = mean_yield_year_dt, aes(x = year, y = mean_yield, color = "Average corn yield across states"), linetype = "dashed") +
#   labs(
#       x = "Year",
#       y = "Corn yield (BU / ACRE)",
#       title = "Corn yield trend by state"
#     ) +
#   scale_color_manual(
#     values = c("Average corn yield across states" = "red")
#     ) +
#   theme_bw() +
#   theme(
#     legend.position = "bottom"
#   )