courtsports

An R package to analyze tennis data and a reproducible workspace.

INSTRUCTIONS

Download our package

devtools::install_github("shannong19/courtsports")

Results

The results of our paper may be reproduced by running the .R/.Rmd files in paper_results. For user convenience, we have provided example code for each of the three sections.

EDA and Visualization

library(courtsports)
library(gridExtra)
library(ggplot2)
data(gs_players)

## Colors
tournament_colors <- c("#0297DB", "#b06835", "#0C2340", "#54008b")
league_colors <- c("#0C2340", "#902BA3")
win_colors <- c("#336699", "#339966")

# with yellow for us
tournament_colors <- c("#0297DB", "#b06835", "#ffe500", "#54008b")

p2 <- ggplot(gs_players) + geom_density(aes(pointswon, color = tournament)) + 
  labs(x = "Points") +
  scale_color_manual("Tournament", values=tournament_colors)

p3 <- ggplot(gs_players) + geom_density(aes(pointswon, color = league))+ 
  labs(x = "Points") +
  scale_color_manual("League", values=league_colors) 




p1 <- ggplot(gs_players) + geom_density(aes(pointswon, color = factor(did_win))) + 
  labs(x = "Points") +
  scale_color_manual("Winner?", values=win_colors) +
  facet_wrap(~league, ncol=1)

grid.arrange(p1, p2, ncol=2,
             top = grid::textGrob("Distribution of points per match",
                            gp=grid::gpar(fontsize=20, fontfamily="serif")),
              layout_matrix = rbind(c(NA, 1),
                                    c(2, 1),
                                    c(NA, 1)),
             heights=c(1, 3, 1))

Hierarchical Modelling

library(courtsports)
library(tidyverse)
library(broom)
library(lme4)

data(gs_players)
data(gs_partial_players)

## Define variables for modeling
gs_players$name_int = as.integer(as.factor(gs_players$name))
gs_players$name_fac = as.factor(gs_players$name)
gs_players$hand_fac = as.factor(gs_players$hand)
gs_players$ioc_fac = as.factor(gs_players$ioc)
gs_players$atp = gs_players$league == "ATP"
gs_players$year_fac = as.factor(gs_players$year)
gs_players$late_round = gs_players$round >= "R16"
gs_players$seeded = gs_players$rank <= 32
gs_players$opponent_seeded = gs_players$opponent_rank <= 32

gs_partial_players$name_int = as.integer(as.factor(gs_partial_players$name))
gs_partial_players$name_fac = as.factor(gs_partial_players$name)
gs_partial_players$ioc_fac = as.factor(gs_partial_players$ioc)
gs_partial_players$atp = gs_partial_players$Tour == "atp"
gs_partial_players$year_fac = as.factor(gs_partial_players$year)
gs_partial_players$late_round = gs_partial_players$round >= "R16"
gs_partial_players$seeded = gs_partial_players$rank <= 32
gs_partial_players$opponent_seeded = gs_partial_players$opponent_rank <= 32

## Fit models used in paper
set.seed(091418)
ind_logistic_noioc = lme4::glmer(did_win ~ late_round + log(rank) + log(opponent_rank) + year_fac + atp + (0 + tournament |name_int), data = gs_players, family = "binomial", nAGQ =0)
n_aces_mod = lme4::lmer(n_aces ~ late_round + log(rank) + log(opponent_rank) + year_fac + atp + (0 + tournament |name_fac), data = gs_partial_players)
n_winners_mod = lme4::lmer(n_winners ~ late_round + log(rank) + log(opponent_rank) + year_fac + atp + (0 + tournament |name_fac), data = gs_partial_players)
n_net_mod = lme4::lmer(n_netpt_w ~ late_round + log(rank) + log(opponent_rank) + year_fac + atp + (0 + tournament |name_fac), data = gs_partial_players)
n_ue_mod = lmer(n_ue ~ late_round + log(rank) + log(opponent_rank) + year_fac + atp + (0 + tournament |name_fac), data = gs_partial_players)

Individual Modelling

library(courtsports)
data(gs_partial_players)

player_name <- "Andy Murray"
ref <- "Wimbledon"
start_lower = TRUE
test_prop <- .50
out_list <- courtsports::model_individual(
                              player_name = player_name,
                              data = gs_partial_players,
                              ref = ref,
                              start_lower = start_lower,
                              test_prop = test_prop)
summary(out_list$final_model)

Data Collection and Transformation

All data is freely available online from Jeff Sackmann's data repositories. In particular, we use S. Kovalchik's deuce R package to interface the repositories. From deuce, we use the following data sets:

Raw data

data("atp_matches", package = "deuce") - Basic results of men's ATP matches including winner, loser, opponent, country of origin, court, and more (See help(atp_matches, package = "deuce"))
data(wta_matches, package = "deuce") - Basic results of women's WTA matches including winner, loser, opponent, country of origin, court, and more (See help(wta_matches, package = "deuce"))
data(gs_point_by_point, package = "deuce") Point by point data of men and women's matches with results on number of winners, unforced errors, net points won, and more. This is not complete (See help(gs_point_by_point, package = "deuce"))

Data Transformation

We transform the raw data to use in our analysis. These complete transformation scripts may be found in the folder data-raw/. We then store the results of these transformations as .rda objects. These 4 primary transformation objects are available in our package for use:

data(gs) - Combined men and women's grand slam matches from 2013-2017
data(gs_players) - Transformation of gs to look at a single player at a time, instead of pairs of opponents
data(gs_partial) - Aggregated point by point data for the grand slams. It is crowd-sourced and partial.
data(gs_partial_players) - Transformation of gs_partial to look at a single player.

The complete and full data description and summaries of the data are available as .Rmd files here.

TROUBLESHOOTING

Make sure courtsports is installed. (devtools::install_github("shannong19/courtsports"))
Use the most recent version and packages in R
Make sure you have the following R packages installed (regular CRAN installation unless otherwise specified)

broom
deuce (install_github("skoval/deuce")
dplyr
forcats
ggplot2
ggpubr
ggrepel
grid
gridExtra
kableExtra
knitr
lme4
MASS
rmarkdown
stringr
tidyr

If none of this works, please feel free to leave a comment.

skgallagher/courtsports