March Madness

Basketball is a multimillion dollar sports that fascinates millions of people all year round peaking in March Madness each year. Here, we are using play-by-play data on all NBA games on two days in Oct 2017.

Question #1: First Overview

Read the data from sample-combined-pbp-stats.csv without downloading the file locally. (Hint: ?read.csv) Each line of the file describes one play in a game.

  • How many rows and columns does the data have? What are the variables called?
  • each individual game is assigned a unique identifier game_id. How many games are recorded in the data and how many plays (number of rows) does a game have on average?
# write your code here

Question #2: Who won?

For each game, the variable points keeps track of the number of points attempted in each play.

  • Use functions from the dplyr package to work out the number of points each team scored (check result). Filter out all plays that are not associated with any of the teams.

  • Compare your scores with the final values of home and away scores (check for event_type and period)

  • Using ggplot2 plot scores by team. Sort teams by their median scores.

  • In a second step, use the scores to identify winning and losing teams of each game (Hint: use dplyr again and remember what which.max and which.min are doing).

  • player keeps track of which player makes a shot. For each game, identify which player in each team scored the most points. (Hint: ?rank)

# write your code here

Question #3 : On the court

The variables converted_x and converted_y give the location of the acting player on the court. Check with the variable event_type to see, for which types of play we have geographic information.

For game number 21700003 plot the geographic location of each play on the court, colour by team and incorporate visually the play’s outcome (variable result).

# write your code here

Question #4: Distance hurts?

Is there any evidence that shots closer to the basket are successful more often? For that, - introduce a new variable d into the data that captures the distance of a player from the basket (basket is in [0,0] for variables original_x and original_y). - draw side-by-side boxplots of distance by result and team (using ggplot2). Interpret. (Hint: you can put either result or team on the x-axis and the remaining variable into facet)

# write your code here

Question #5: Timeline

The variable elapsed is recorded in hour-minute-second format. Convert the information into seconds (hint: you could introduce helper variables for the conversion). The elapsed time variable starts over in each period (a period has 12 mins). Introduce a new variable called time_played that keeps track of the time (in seconds) from the beginning of a game to the end.

Plot the timeline (time_played) of events (event_type) for game with the id 21700002 in a scatterplot. Colour by team. Make Facet by period. Comment on the result.

# write your code here

Question #6: Who is playing?

Variables h1 through h5 and a1 through a5 are the five players of the home team and the away team on the field at that moment in positions 1 through 5. Reshape the data set to combine all player variables. Extract position numbers. For each game identify how many players were playing in each position.

Draw side-by-side boxplots of the number of players by position. Comment.

# write your code here