ncaahoopR <img src="figures/logo.png" align="right" />

ncaahoopR is an R package for working with NCAA Basketball Play-by-Play Data. It scrapes play-by-play data and returns it to the user in a tidy format, allowing the user to explore the data with assist networks, shot charts, and in-game win-probability charts.

For pre-scraped schedules, rosters, box scores, and play-by-play data, check out the ncaahoopR_data repository.

To see the lastest changes in version 1.5, view the change log here.

Installation

You can install ncaahoopR from GitHub with:

# install.packages("devtools")
devtools::install_github("lbenz730/ncaahoopR")

If you encounter installation issues, the following tips have helped a few users successfully install the package:

• If given the option to compile any packages from source rather than installing existing binaries, choose 'No'.
• Windows users with trouble installing the package should try running the following command before reinstalling the package: Sys.setenv(R_REMOTES_NO_ERRORS_FROM_WARNINGS = "true")
• Windows users with trouble installing devtools should try first installing the backports package via install.packages("backports").

Functions

Several functions use ESPN game_ids. You can find the game_id in the URL for the game summary, as shown in the URL for the summary of the UMBC-Virginia game below.

Scraping Data

• get_pbp(team, season): Get entire current season's worth of play-by-play data for a given team and season. season defaults to current season, but can be specified in "2019-20" form.

• get_pbp_game(game_ids, extra_parse): Get play-by-play data for a specific vector of ESPN game_ids. extra_parse is a logical whether to link shot variables and attempt possesion parsing. Default = TRUE.

• get_roster(team, season): Get a particular team's roster. season defaults to current season, but can be specified in "2019-20" form.

• get_schedule(team, season): Get a team's schedule. season defaults to current season, but can be specified in "2019-20" form.

• get_game_ids(team, season): Get a vector of ESPN game_ids for all games involving team specified. season defaults to current season, but can be specified in "2019-20" form.

• get_master_schedule(date): Get schedule of all games for given date. Use YYYY-MM-DD date formatting.

• get_boxscore(game_id): Returns list of 2 data frames, one with each teams' box score for the game in question. Written by Jared Andrews.

• season_boxscore(team, season = current_season, aggregate = 'average'): Returns (aggregated) player stats over the course of a season for a given team. * team: team to return player stats for. * season: of form YYYY-YY. Defaults to current season. * aggregate: one of 'average' (per-game average statistics), 'total' (sums of season stats) or 'raw' (just return all box scores binded together). 'average' is the default. Contributed in collaboration with Kurt Wirth

The team parameter in the above functions must be a valid team name from the ids dataset built into the package. See the Datasets section below for more details.

Win-Probability and Game-Flow Charts

Win Probability Charts

The latest function for plotting win probability charts is wp_chart_new. Following the 2021-22 season other win probability chart functions will be deprecated and replaced by this function (it will be renamed to wp_chart but I don't want to break any existing pipelines during the season). It no longer requires users to input colors. For best results consider saving via ggsave(filename, height = 9/1.2, width = 16/1.2) (or some other 16/9 aspect ratio.)

wp_chart_new(game_id, home_col = NULL, away_col = NULL, include_spread = T, show_legend = T)

• game_id ESPN game_id for the desired contest.
• home_col Chart color for home team (if NULL will default to ncaa_colors primary_color field).
• away_col: Chart color for away team (if NULL will default to ncaa_colors primary_color field).
• include_spread: Logical, whether to include pre-game spread in Win Probability calculations. Default = TRUE.
• show_legend: Logical, whether or not to show legend/text on chart. Default = TRUE.

A prior version of wp_chart used base R while gg_wp_chart used the ggplot2 plotting library. As of the 2020-21 season, both functions call the same ggplot2 library, and gg_wp_chart now simply aliases wp_chart

wp_chart(game_id, home_col, away_col, include_spread = T, show_legend = T)

• game_id ESPN game_id for the desired contest.
• home_col Chart color for home team.
• away_col: Chart color for away team.
• include_spread: Logical, whether to include pre-game spread in Win Probability calculations. Default = TRUE.
• show_legend: Logical, whether or not to show legend/text on chart. Default = TRUE.

gg_wp_chart(game_id, home_col, away_col, show_labels = T)

• game_id ESPN game_id for the desired contest.
• home_col Chart color for home team.
• away_col: Chart color for away team.
• include_spread: Logical, whether to include pre-game spread in Win Probability calculations. Default = TRUE.
• show_labels: Logical whether Game Excitement Index and Minimum Win Probability metrics should be displayed on the plot. Default = TRUE.

Game Flow Charts

game_flow(game_id, home_col, away_col)

• game_id ESPN game_id for the desired contest.
• home_col Chart color for home team.
• away_col Chart color for away team.

Game Excitement Index

game_exciment_index(game_id, include_spread = T)

• include_spread: Logical, whether to include pre-game spread in Win Probability calculations. Default = TRUE.

Returns GEI (Game Excitement Index) for given ESPN game_id. For more information about how these win-probability charts are fit and how Game Excitement Index is calculated, check out the below links

• Model Methodology
• Game Excitement Index

Game Control Measures

average_win_prob(game_id, include_spread = T)

• ESPN game_id for which to compute time-based average win probability (from perspective of home team).
• include_spread: Logical, whether to include pre-game spread in Win Probability calculations. Default = TRUE.

average_score_diff(game_id)

• ESPN game_id for which to compute time-based average score differential (from perspective of home team).

Assist Networks

Traditional Assist Networks

assist_net(team, season, node_col, three_weights = T, threshold = T, message = NA, return_stats = T)

• team is the ESPN team name, as listed in the ids data frame.
• season Options include "2018-19" (for entire season), or a vector of ESPN game IDs.
• node_col is the node color for the graph.
• three_weights (default = TRUE): Logical. If TRUE, assisted three-point shots are given a weight of 1.5. If FALSE, assisted three-point shots are given a weight of 1. In both cases, assisted two-point shots are given a weight of 1.
• threshold (default = 0) Number between 0-1 indicating minimum percentage of team's assisted baskets a player needs to be involved in to be included in network graph.
• message (default = NA) Option for custom message to replace graph title when using a subset of the season (e.g. conference play).
• return_stats (default = TRUE) Return Assist Network-related statistics

Circle Assist Networks and Player Highlighting

circle_assist_net(team, season, highlight_player = NA, highlight_color = NA, three_weights = T, threshold = 0, message = NA, return_stats = T)

• team is the ESPN team name, as listed in the ids data frame.
• season: Options include "YYYY-YY" (for entire season), or a vector of ESPN game IDs.
• highlight_player (default = NA) Name of player to highlight in assist network. NA yields full-team assist network with no player highlighting.
• highlight_color (default = NA) Color of player links to be highlighted. NA if highlight_player is NA.
• three_weights (default = TRUE): Logical. If TRUE, assisted three-point shots are given a weight of 1.5. If FALSE, assisted three-point shots are given a weight of 1. In both cases, assisted two-point shots are given a weight of 1.
• threshold (default = 0) Number between 0-1 indicating minimum percentage of team's assisted baskets a player needs to be involved in to be included in network graph.
• message (default = NA) User-supplied plot title to overwrite default plot title, if desired.
• return_stats (default = TRUE) Return Assist Network-related statistics

Shot Charts

There are currently three functions for scraping and plotting shot location data. These functions are written by Meyappan Subbaiah.

get_shot_locs(game_id): Returns data frame with shot location data when available. Note that if the extra_parse flag in get_pbp_game is set to TRUE, shot location data will already be included in the play-by-play data (if available).

• game_id: ESPN game_id from which shot locations should be scraped.

game_shot_chart(game_id, heatmap = F): Plots shots for a given game.

• game_id: ESPN game_id from which shot locations should be scraped.
• heatmap (default = FALSE): Logical, whether to use density-heat map or plot individual points.
• shot-plotting colors derived from team's primary color listed in ncaa_colors data frame.

team_shot_chart(game_ids, team, heatmap = F): Plots shots taken by team during a given set of g