Lesson 6 Assignment Solutions

library(dplyr)
library(tidyr)
library(reactable)
library(plotly)
library(crosstalk)

The Dataset

Read in the dataset from the Lesson 5 Assignment and then quickly check the first few rows (via head from base R, slice_head from dplyr, reactable, DT, etc.) of this dataset to preview what’s in it.

df <- read.csv("https://manctsui.github.io/datasets/timesData.csv")
reactable(df, defaultPageSize = 2, searchable = TRUE)

Problem 1A

Find the three countries with the most universities on the above rankings.

Display this information as a table (reactable, DT, etc.) with two columns:

• The first column is the country name
• The second column is the number of times the country’s universities featured in the above rankings.

# Summarize/aggregate by country, sort, slice head
by_country <- df |>
    group_by(country) |>
    count() |>
    arrange(desc(n)) |>
    head(3)

# Rename columns and display resulting table as a reactable.
by_country |>
    select("Country" = country, "No. of Ranked Universities" = n) |>
    reactable()

Problem 1B

Display the result of Problem 1A as a barchart in ggplot or plotly.

fig <- plot_ly(by_country, x = ~country, y = ~n, type="bar") |>
    layout(
        yaxis = list(title = "No. of Ranked Universities"),
        xaxis = list(title = "Country")
        )

fig

Problem 2A

Display the top 50 universities in the year 2011 as a table with only two columns:

• First column is the rank.
• Second column is the university name.

Answer: Use a spreadsheet program (Microsoft Excel, etc.) on our CSV file or use df |> filter(year == 2011) |> distinct(world_rank) to see that the world rankings are character strings of numbers. So we just need to convert these character strings to numerics via as.numeric() function from base R.

(Note: in other years, world rankings include strings like “201-225” or “=39” which require more work to clean.)

top <- df |>
    filter(year == 2011) |>
    arrange(as.numeric(world_rank)) |>
    select("Rank" = world_rank, "University" = university_name) |>
    head(50)

top$Rank <- as.numeric(top$Rank)

top |>
    reactable()

Problem 2B

If we are to turn the data frame that results from Problem 1A as a SharedData object, would $rank column be a good candidate to use as a key? Explain.

Answer: Two universities may share the same rank, so rank alone cannot distinguish between two rows/observations.

Problem 2C

Create a filter slider from crosstalk that allows us to narrow our selection of universities based on the rank of the university.

sd <- SharedData$new(top, key = ~University)
filter_slider("slider", "Rank", sd, ~Rank)

Rank

reactable(sd)

Problem 3

Do two separate scatterplots from our university world ranking dataset subsetted to 2016:

• First scatterplot is number of students vs. student-staff ratio. (Do you see any visible outlier?)
• Second scatterplot is student-staff ratio vs. teaching score.
• Link the two scatterplots together via crosstalk.

df2 <- df |>
    filter(year == 2016)

sd2 <- SharedData$new(df2, key = ~university_name)

scatter1 <- plot_ly(sd2, x = ~num_students, y = ~student_staff_ratio, type = "scatter", mode = "markers") |>
  add_trace(
    text = df2[["university_name"]],
    hoverinfo = 'text',
    showlegend = FALSE
  ) |>
  layout(
        xaxis = list(title = "Number of Students"),
        yaxis = list(title = "Student-to-Staff Ratio")
  )
scatter2 <- plot_ly(sd2, x = ~student_staff_ratio, y = ~teaching, type = "scatter", mode = "markers") |>
  add_trace(
    text = df2$university_name,
    hoverinfo = 'text',
    showlegend = FALSE
  ) |>
  layout(
        xaxis = list(title = "Student-to-Staff Ratio"),
        yaxis = list(title = "Teaching Score")
  )

scatter1

Warning: Ignoring 7 observations
Warning: Ignoring 7 observations

scatter2

Warning: Ignoring 7 observations
Warning: Ignoring 7 observations

We added hover text showing the university name via the add_trace(...) option by following plotly’s hover text documentation. You do not need to do this.