, Paola Zuccolotto
[aut] , Marica
Manisera [aut] | Title: | Analysis and Visualization of Basketball Data |
|---|---|
| Description: | Contains data and code to accompany the book P. Zuccolotto and M. Manisera (2020) Basketball Data Science. Applications with R. CRC Press. ISBN 9781138600799. |
| Authors: | Marco Sandri [aut, cre]
, Paola Zuccolotto
[aut] , Marica
Manisera [aut] |
| Maintainer: | Marco Sandri <[email protected]> |
| License: | GPL (>= 2.0) |
| Version: | 0.7.0 |
| Built: | 2025-02-17 05:31:56 UTC |
| Source: | https://github.com/sndmrc/basketballanalyzer |
The assistnet command provides a comprehensive analysis of a team's assist-shot network, revealing crucial insights into player interactions and on-court dynamics.
assistnet( data, assist = "assist", player = "player", points = "points", event.type = "event_type", normalize = FALSE, period.length = 12, time.thr = 0 )assistnet( data, assist = "assist", player = "player", points = "points", event.type = "event_type", normalize = FALSE, period.length = 12, time.thr = 0 )
data |
a data frame whose rows are field shots and columns are variables to be specified in |
assist |
character, indicating the name of the variable with players who made the assists, if any. |
player |
character, indicating the name of the variable with players who made the shot. |
points |
character, indicating the name of the variable with points. |
event.type |
character, indicating the name of the variable with type of event (mandatory categories are |
normalize |
logical, if |
period.length |
numerical, the length of a quarter in minutes (default: 12 minutes as in NBA) |
time.thr |
numerical, (default |
The data data frame could also be a play-by-play dataset provided that rows corresponding to events different from field shots are not coded as "shot" in the event.type variable. (To be completed)
Normalization: \[4 \cdot \text{(period.length)} \cdot \frac{(\text{number of assists})}{\text{(minutes played in attack by each couple of players)}}\]
A list with 3 elements, assistTable (a table), nodeStats (a data frame), and assistNet (a network object). See Details.
assistTable, the cross-table of assists made and received by the players.
nodeStats, a data frame with the following variables:
FGM (fields goals made),
FGM_AST (field goals made thanks to a teammate's assist),
FGM_ASTp (percentage of FGM_AST over FGM),
FGPTS (points scored with field goals),
FGPTS_AST (points scored thanks to a teammate's assist),
FGPTS_ASTp (percentage of FGPTS_AST over FGPTS),
AST (assists made),
ASTPTS (point scored by assist's teammates).
minTable (da completare)
assistminTable (da completare)
assistNet, an object of class network that can be used for further network analysis with specific R packages (see network)
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW") out <- assistnet(PbP.GSW)PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW") out <- assistnet(PbP.GSW)
Draws a bar-line plot
barline( data, id, bars, line, order.by = id, decreasing = TRUE, labels.bars = NULL, label.line = NULL, position.bars = "stack", title = NULL )barline( data, id, bars, line, order.by = id, decreasing = TRUE, labels.bars = NULL, label.line = NULL, position.bars = "stack", title = NULL )
data |
a data frame. |
id |
character, name of the ID variable. |
bars |
character vector, names of the bar variables. |
line |
character, name of the line variable. |
order.by |
character, name of the variable used to order bars (on the x-axis). |
decreasing |
logical; if |
labels.bars |
character vector, labels for the bar variables. |
label.line |
character, label for the line variable on the second y-axis (on the right). |
position.bars |
character, used to adjust the positioning of the bars in the plot; there are four main options: |
title |
character, plot title. |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
dts <- subset(Pbox, Team=="Houston Rockets" & MIN>=500) barline(data=dts, id="Player", bars=c("P2p","P3p","FTp"), line="MIN", order.by="Player", labels.bars=c("2P","3P","FT"), title="Houston Rockets")dts <- subset(Pbox, Team=="Houston Rockets" & MIN>=500) barline(data=dts, id="Player", bars=c("P2p","P3p","FTp"), line="MIN", order.by="Player", labels.bars=c("2P","3P","FT"), title="Houston Rockets")
Draws a bubble plot
bubbleplot( data, id, x, y, col, size, text.col = NULL, text.size = 2.5, scale.size = TRUE, labels = NULL, mx = NULL, my = NULL, mcol = NULL, title = NULL, repel = TRUE, text.legend = TRUE, hline = TRUE, vline = TRUE )bubbleplot( data, id, x, y, col, size, text.col = NULL, text.size = 2.5, scale.size = TRUE, labels = NULL, mx = NULL, my = NULL, mcol = NULL, title = NULL, repel = TRUE, text.legend = TRUE, hline = TRUE, vline = TRUE )
data |
a data frame. |
id |
character, name of the ID variable. |
x |
character, name of the x-axis variable. |
y |
character, name of the y-axis variable. |
col |
character, name of variable on the color axis. |
size |
character, name of variable on the size axis. |
text.col |
character, name of variable for text colors. |
text.size |
numeric, text font size (default 2.5). |
scale.size |
logical; if |
labels |
character vector, variable labels (on legend and axis). |
mx |
numeric, x-coordinate of the vertical axis; default is the mean value of |
my |
numeric, y-coordinate of the horizontal axis; default is the mean value of |
mcol |
numeric, midpoint of the diverging scale (see |
title |
character, plot title. |
repel |
logical; if |
text.legend |
logical; if |
hline |
logical; if |
vline |
logical; if |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
X <- with(Tbox, data.frame(T=Team, P2p=P2p, P3p=P3p, FTp=FTp, AS=P2A+P3A+FTA)) labs <- c("2-point shots (% made)","3-point shots (% made)", "free throws (% made)","Total shots attempted") bubbleplot(X, id="T", x="P2p", y="P3p", col="FTp", size="AS", labels=labs)X <- with(Tbox, data.frame(T=Team, P2p=P2p, P3p=P3p, FTp=FTp, AS=P2A+P3A+FTA)) labs <- c("2-point shots (% made)","3-point shots (% made)", "free throws (% made)","Total shots attempted") bubbleplot(X, id="T", x="P2p", y="P3p", col="FTp", size="AS", labels=labs)
Correlation analysis
corranalysis(data, threshold = 0, sig.level = 0.95)corranalysis(data, threshold = 0, sig.level = 0.95)
data |
a numeric matrix or data frame (see |
threshold |
numeric, correlation cutoff (default 0); correlations in absolute value below |
sig.level |
numeric, significance level (default 0.95); correlations with p-values greater that |
A list with the following elements:
corr.mtx (the complete correlation matrix)
corr.mtx.trunc (the truncated correlation matrix)
cor.mtest (the output of the significance test on correlations; see cor.mtest)
threshold correlation cutoff
sig.level significance level
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- data.frame(Pbox$PTS,Pbox$P3M,Pbox$P2M, Pbox$OREB + Pbox$DREB,Pbox$AST, Pbox$TOV,Pbox$STL,Pbox$BLK)/Pbox$MIN names(data) <- c("PTS","P3M","P2M","REB","AST","TOV","STL","BLK") data <- subset(data, Pbox$MIN >= 500) out <- corranalysis(data, threshold = 0.5)data <- data.frame(Pbox$PTS,Pbox$P3M,Pbox$P2M, Pbox$OREB + Pbox$DREB,Pbox$AST, Pbox$TOV,Pbox$STL,Pbox$BLK)/Pbox$MIN names(data) <- c("PTS","P3M","P2M","REB","AST","TOV","STL","BLK") data <- subset(data, Pbox$MIN >= 500) out <- corranalysis(data, threshold = 0.5)
R function CreateRadialPlot by William D. Vickers, freely downloadable from the web
CreateRadialPlot( plot.data, axis.labels = colnames(plot.data)[-1], grid.min = -0.5, grid.mid = 0, grid.max = 0.5, centre.y = grid.min - ((1/9) * (grid.max - grid.min)), plot.extent.x.sf = 1.2, plot.extent.y.sf = 1.2, x.centre.range = 0.02 * (grid.max - centre.y), label.centre.y = FALSE, grid.line.width = 0.5, gridline.min.linetype = "longdash", gridline.mid.linetype = "longdash", gridline.max.linetype = "longdash", gridline.min.colour = "grey", gridline.mid.colour = "blue", gridline.max.colour = "grey", grid.label.size = 4, gridline.label.offset = -0.02 * (grid.max - centre.y), label.gridline.min = TRUE, axis.label.offset = 1.15, axis.label.size = 2.5, axis.line.colour = "grey", group.line.width = 1, group.point.size = 4, background.circle.colour = "yellow", background.circle.transparency = 0.2, plot.legend = if (nrow(plot.data) > 1) TRUE else FALSE, legend.title = "Player", legend.text.size = grid.label.size, titolo = FALSE )CreateRadialPlot( plot.data, axis.labels = colnames(plot.data)[-1], grid.min = -0.5, grid.mid = 0, grid.max = 0.5, centre.y = grid.min - ((1/9) * (grid.max - grid.min)), plot.extent.x.sf = 1.2, plot.extent.y.sf = 1.2, x.centre.range = 0.02 * (grid.max - centre.y), label.centre.y = FALSE, grid.line.width = 0.5, gridline.min.linetype = "longdash", gridline.mid.linetype = "longdash", gridline.max.linetype = "longdash", gridline.min.colour = "grey", gridline.mid.colour = "blue", gridline.max.colour = "grey", grid.label.size = 4, gridline.label.offset = -0.02 * (grid.max - centre.y), label.gridline.min = TRUE, axis.label.offset = 1.15, axis.label.size = 2.5, axis.line.colour = "grey", group.line.width = 1, group.point.size = 4, background.circle.colour = "yellow", background.circle.transparency = 0.2, plot.legend = if (nrow(plot.data) > 1) TRUE else FALSE, legend.title = "Player", legend.text.size = grid.label.size, titolo = FALSE )
plot.data |
plot.data |
axis.labels |
axis.labels |
grid.min |
grid.min |
grid.mid |
grid.mid |
grid.max |
grid.max |
centre.y |
centre.y |
plot.extent.x.sf |
plot.extent.x.sf |
plot.extent.y.sf |
plot.extent.y.sf |
x.centre.range |
x.centre.range |
label.centre.y |
label.centre.y |
grid.line.width |
grid.line.width |
gridline.min.linetype |
gridline.min.linetype |
gridline.mid.linetype |
gridline.mid.linetype |
gridline.max.linetype |
gridline.max.linetype |
gridline.min.colour |
gridline.min.colour |
gridline.mid.colour |
gridline.mid.colour |
gridline.max.colour |
gridline.max.colour |
grid.label.size |
grid.label.size |
gridline.label.offset |
gridline.label.offset |
label.gridline.min |
label.gridline.min |
axis.label.offset |
axis.label.offset |
axis.label.size |
axis.label.size |
axis.line.colour |
axis.line.colour |
group.line.width |
group.line.width |
group.point.size |
group.point.size |
background.circle.colour |
background.circle.colour |
background.circle.transparency |
background.circle.transparency |
plot.legend |
plot.legend |
legend.title |
legend.title |
legend.text.size |
legend.text.size |
titolo |
plot title |
A description of the function can be found at the following link: http://rstudio-pubs-static.s3.amazonaws.com/5795_e6e6411731bb4f1b9cc7eb49499c2082.html
Vickers D.W. (2006) Multi-Level Integrated Classifications Based on the 2001 Census, PhD Thesis, School of Geography, The University of Leeds
Computes and plots kernel density estimation of shots with respect to a concurrent variable
densityplot( data, var, shot.type = "field", thresholds = NULL, best.scorer = FALSE, period.length = 12, bw = NULL, title = NULL )densityplot( data, var, shot.type = "field", thresholds = NULL, best.scorer = FALSE, period.length = 12, bw = NULL, title = NULL )
data |
a data frame whose rows are shots and with the following columns: |
var |
character, a string giving the name of the numerical variable according to which the shot density is estimated. Available options: |
shot.type |
character, a string giving the type of shots to be analyzed. Available options: |
thresholds |
numerical vector with two thresholds defining the range boundaries that divide the area under the density curve into three regions. If |
best.scorer |
logical; if TRUE, displays the player who scored the highest number of points in the corresponding interval. |
period.length |
numeric, the length of a quarter in minutes (default: 12 minutes as in NBA). |
bw |
numeric, the value for the smoothing bandwidth of the kernel density estimator or a character string giving a rule to choose the bandwidth (see density). |
title |
character, plot title. |
The data data frame could also be a play-by-play dataset provided that rows corresponding to events different from shots have NA in the ShotType variable.
Required columns:
ShotType, a factor with the following levels: "2P", "3P", "FT" (and NA for events different from shots)
player, a factor with the name of the player who made the shot
points, a numeric variable (integer) with the points scored by made shots and 0 for missed shots
playlength, a numeric variable with time between the shot and the immediately preceding event
periodTime, a numeric variable with seconds played in the quarter when the shot is attempted
totalTime, a numeric variable with seconds played in the whole match when the shot is attempted
shot_distance, a numeric variable with the distance of the shooting player from the basket (in feet)
A ggplot2 plot
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) data.team <- subset(PbP, team=="GSW" & result!="") densityplot(data=data.team, shot.type="2P", var="playlength", best.scorer=TRUE) data.opp <- subset(PbP, team!="GSW" & result!="") densityplot(data=data.opp, shot.type="2P", var="shot_distance", best.scorer=TRUE)PbP <- PbPmanipulation(PbP.BDB) data.team <- subset(PbP, team=="GSW" & result!="") densityplot(data=data.team, shot.type="2P", var="playlength", best.scorer=TRUE) data.opp <- subset(PbP, team!="GSW" & result!="") densityplot(data=data.opp, shot.type="2P", var="shot_distance", best.scorer=TRUE)
Add lines of NBA court to an existing ggplot2 plot
drawNBAcourt(p, size = 1.5, col = "black", full = FALSE)drawNBAcourt(p, size = 1.5, col = "black", full = FALSE)
p |
a ggplot2 object. |
size |
numeric, line size. |
col |
line color. |
full |
logical; if TRUE draws a complete NBA court; if FALSE draws a half court. |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
library(ggplot2) p <- ggplot(data.frame(x=0, y=0), aes(x,y)) + coord_fixed() drawNBAcourt(p)library(ggplot2) p <- ggplot(data.frame(x=0, y=0), aes(x,y)) + coord_fixed() drawNBAcourt(p)
Plots expected points of shots as a function of the distance from the basket (default) or another variable
expectedpts( data, var = "shot_distance", players = NULL, bw = 10, period.length = 12, palette = gg_color_hue, team = TRUE, col.team = "gray", col.hline = "black", xlab = NULL, x.range = "auto", title = NULL, legend = TRUE )expectedpts( data, var = "shot_distance", players = NULL, bw = 10, period.length = 12, palette = gg_color_hue, team = TRUE, col.team = "gray", col.hline = "black", xlab = NULL, x.range = "auto", title = NULL, legend = TRUE )
data |
a data frame whose rows are field shots and with the following columns: |
var |
character, a string giving the name of the numerical variable according to which the expected points are estimated; available options |
players |
subset of players to be displayed (optional; it can be used only if the |
bw |
numeric, smoothing bandwidth of the kernel density estimator (see |
period.length |
numeric, the length of a quarter in minutes (default: 12 minutes as in NBA). |
palette |
color palette. |
team |
logical; if |
col.team |
character, color of the expected points line for all the shots in data (default |
col.hline |
character, color of the dashed horizontal line (default |
xlab |
character, x-axis label. |
x.range |
numerical vector or character; available options: |
title |
character, plot title. |
legend |
logical, if |
The data data frame could also be a play-by-play dataset provided that rows corresponding to events different from field shots have values different from "shot" or "miss" in the even_type variable.
Required columns:
event_type, a factor with the following levels: "shot" for made field shots and "miss" for missed field shots
player, a factor with the name of the player who made the shot
points, a numeric variable (integer) with the points scored by made shots and 0 for missed shots
playlength, a numeric variable with time between the shot and the immediately preceding event
periodTime, a numeric variable with seconds played in the quarter when the shot is attempted
totalTime, a numeric variable with seconds played in the whole match when the shot is attempted
shot_distance, a numeric variable with the distance of the shooting player from the basket (in feet)
A ggplot2 plot
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & !is.na(shot_distance)) plrys <- c("Stephen Curry","Kevin Durant") expectedpts(data=PbP.GSW, bw=10, players=plrys, col.team='dodgerblue', palette=colorRampPalette(c("gray","black")), col.hline="red")PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & !is.na(shot_distance)) plrys <- c("Stephen Curry","Kevin Durant") expectedpts(data=PbP.GSW, bw=10, players=plrys, col.team='dodgerblue', palette=colorRampPalette(c("gray","black")), col.hline="red")
Calculates possessions, pace, offensive and defensive rating, and Four Factors
fourfactors(TEAM, OPP)fourfactors(TEAM, OPP)
TEAM |
a data frame whose rows are the analyzed teams and with columns referred to the team achievements in the considered games (a box score); required variables: |
OPP |
a data frame whose rows are the analyzed teams and with columns referred to the achievements of the opponents of each team in the considered game; required variables: |
The rows of the TEAM and the OPP data frames must be referred to the same teams in the same order.
Required columms:
Team, a factor with the name of the analyzed team
P2A, a numeric variable (integer) with the number of 2-points shots attempted
P2M, a numeric variable (integer) with the number of 2-points shots made
P3A, a numeric variable (integer) with the number of 3-points shots attempted
P3M, a numeric variable (integer) with the number of 3-points shots made
FTA, a numeric variable (integer) with the number of free throws attempted
FTM, a numeric variable (integer) with the number of free throws made
OREB, a numeric variable (integer) with the number of offensive rebounds
DREB, a numeric variable (integer) with the number of defensive rebounds
TOV, a numeric variable (integer) with the number of turnovers
MIN, a numeric variable (integer) with the number of minutes played
An object of class fourfactors, i.e. a data frame with the following columns:
Team, a factor with the name of the analyzed team
POSS.Off, a numeric variable with the number of possessions of each team calculated with the formula
POSS.Def, a numeric variable with the number of possessions of the opponents of each team calculated with the formula
PACE.Off, a numeric variable with the pace of each team (number of possessions per minute played)
PACE.Def, a numeric variable with the pace of the opponents of each team (number of possessions per minute played)
ORtg, a numeric variable with the offensive rating (the points scored by each team per 100 possessions)
DRtg, a numeric variable with the defensive rating (the points scored by the opponents of each team per 100 possessions)
F1.Off, a numeric variable with the offensive first factor (effective field goal percentage)
F2.Off, a numeric variable with the offensive second factor (turnovers per possession)
F3.Off, a numeric variable with the offensive third factor (rebouding percentage)
F4.Off, a numeric variable with the offensive fourth factor (free throw rate)
F1.Def, a numeric variable with the defensive first factor (effective field goal percentage)
F2.Def, a numeric variable with the defensive second factor (turnovers per possession)
F3.Def, a numeric variable with the defensive third factor (rebouding percentage)
F4.Def, a numeric variable with the defensive fourth factor (free throw rate)
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
selTeams <- c(2,6,10,11) FF <- fourfactors(Tbox[selTeams,], Obox[selTeams,]) plot(FF)selTeams <- c(2,6,10,11) FF <- fourfactors(Tbox[selTeams,], Obox[selTeams,]) plot(FF)
Agglomerative hierarchical clustering
hclustering(data, k = NULL, nclumax = 10, labels = NULL, linkage = "ward.D")hclustering(data, k = NULL, nclumax = 10, labels = NULL, linkage = "ward.D")
data |
numeric data frame. |
k |
integer, number of clusters. |
nclumax |
integer, maximum number of clusters (when |
labels |
character, row labels. |
linkage |
character, the agglomeration method to be used in |
The hclustering function performs a preliminary standardization of columns in data.
A hclustering object.
If k is NULL, the hclustering object is a list of 3 elements:
k NULL
clusterRange integer vector, values of k (from 1 to nclumax) at which the variance between of the clusterization is evaluated
VarianceBetween numeric vector, values of the variance between evaluated for k in clusterRange
If k is not NULL, the hclustering object is a list of 5 elements:
k integer, number of clusters
Subjects data frame, subjects' cluster identifiers
ClusterList list, clusters' composition
Profiles data frame, clusters' profiles, i.e. the average of the variables within clusters and the cluster eterogeineity index (CHI)
Hclust an object of class hclust, see hclust
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- with(Pbox, data.frame(PTS, P3M, REB=OREB+DREB, AST, TOV, STL, BLK, PF)) data <- subset(data, Pbox$MIN >= 1500) ID <- Pbox$Player[Pbox$MIN >= 1500] hclu1 <- hclustering(data) plot(hclu1) hclu2 <- hclustering(data, labels=ID, k=7) plot(hclu2)data <- with(Pbox, data.frame(PTS, P3M, REB=OREB+DREB, AST, TOV, STL, BLK, PF)) data <- subset(data, Pbox$MIN >= 1500) ID <- Pbox$Player[Pbox$MIN >= 1500] hclu1 <- hclustering(data) plot(hclu1) hclu2 <- hclustering(data, labels=ID, k=7) plot(hclu2)
Inequality analysis
inequality(data, nplayers)inequality(data, nplayers)
data |
numeric vector containing the achievements (e.g. scored points) of the players whose inequality has to be analyzed. |
nplayers |
integer, number of players to include in the analysis (ranked in nondecreasing order according to the values in data). |
A list with the following elements: Lorenz (cumulative distributions used to plot the Lorenz curve) and Gini (Gini coefficient).
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.BN <- subset(Pbox, Team=="Brooklyn Nets") out <- inequality(Pbox.BN$PTS, nplayers=8) print(out) plot(out)Pbox.BN <- subset(Pbox, Team=="Brooklyn Nets") out <- inequality(Pbox.BN$PTS, nplayers=8) print(out) plot(out)
Reports whether x is a 'networkdata' object
is.assistnet(x)is.assistnet(x)
x |
an object to test. |
Returns TRUE if its argument is of class networkdata and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & player!="") out <- assistnet(PbP.GSW) is.assistnet(out)PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & player!="") out <- assistnet(PbP.GSW) is.assistnet(out)
Reports whether x is a 'corranalysis' object
is.corranalysis(x)is.corranalysis(x)
x |
an object to test. |
Returns TRUE if its argument is of class corranalysis and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- data.frame(Pbox$PTS,Pbox$P3M,Pbox$P2M, Pbox$OREB + Pbox$DREB,Pbox$AST, Pbox$TOV,Pbox$STL,Pbox$BLK)/Pbox$MIN names(data) <- c("PTS","P3M","P2M","REB","AST","TOV","STL","BLK") data <- subset(data, Pbox$MIN >= 500) out <- corranalysis(data) is.corranalysis(out)data <- data.frame(Pbox$PTS,Pbox$P3M,Pbox$P2M, Pbox$OREB + Pbox$DREB,Pbox$AST, Pbox$TOV,Pbox$STL,Pbox$BLK)/Pbox$MIN names(data) <- c("PTS","P3M","P2M","REB","AST","TOV","STL","BLK") data <- subset(data, Pbox$MIN >= 500) out <- corranalysis(data) is.corranalysis(out)
Reports whether x is a 'fourfactors' object
is.fourfactors(x)is.fourfactors(x)
x |
an object to test. |
Returns TRUE if its argument is of class fourfactors and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
selTeams <- c(2,6,10,11) out <- fourfactors(Tbox[selTeams,], Obox[selTeams,]) is.fourfactors(out)selTeams <- c(2,6,10,11) out <- fourfactors(Tbox[selTeams,], Obox[selTeams,]) is.fourfactors(out)
Reports whether x is a 'hclustering' object
is.hclustering(x)is.hclustering(x)
x |
an object to test. |
Returns TRUE if its argument is of class hclustering and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- data.frame(Pbox$PTS,Pbox$P3M, Pbox$OREB + Pbox$DREB, Pbox$AST, Pbox$TOV, Pbox$STL, Pbox$BLK,Pbox$PF) names(data) <- c("PTS","P3M","REB","AST","TOV","STL","BLK","PF") data <- subset(data, Pbox$MIN >= 1500) ID <- Pbox$Player[Pbox$MIN >= 1500] hclu <- hclustering(data, labels=ID, k=7) is.hclustering(hclu)data <- data.frame(Pbox$PTS,Pbox$P3M, Pbox$OREB + Pbox$DREB, Pbox$AST, Pbox$TOV, Pbox$STL, Pbox$BLK,Pbox$PF) names(data) <- c("PTS","P3M","REB","AST","TOV","STL","BLK","PF") data <- subset(data, Pbox$MIN >= 1500) ID <- Pbox$Player[Pbox$MIN >= 1500] hclu <- hclustering(data, labels=ID, k=7) is.hclustering(hclu)
Reports whether x is a 'inequality' object.
is.inequality(x)is.inequality(x)
x |
an object to test. |
Returns TRUE if its argument is of class inequality and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.BN <- subset(Pbox, Team=="Brooklyn Nets") out <- inequality(Pbox.BN$PTS, npl=8) is.inequality(out)Pbox.BN <- subset(Pbox, Team=="Brooklyn Nets") out <- inequality(Pbox.BN$PTS, npl=8) is.inequality(out)
Reports whether x is a 'kclustering' object
is.kclustering(x)is.kclustering(x)
x |
an object to test. |
Returns TRUE if its argument is of class kclustering and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
FF <- fourfactors(Tbox,Obox) X <- with(FF, data.frame(OD.Rtg=ORtg/DRtg, F1.r=F1.Def/F1.Off, F2.r=F2.Off/F2.Def, F3.O=F3.Def, F3.D=F3.Off)) X$P3M <- Tbox$P3M X$STL.r <- Tbox$STL/Obox$STL kclu <- kclustering(X) is.kclustering(kclu)FF <- fourfactors(Tbox,Obox) X <- with(FF, data.frame(OD.Rtg=ORtg/DRtg, F1.r=F1.Def/F1.Off, F2.r=F2.Off/F2.Def, F3.O=F3.Def, F3.D=F3.Off)) X$P3M <- Tbox$P3M X$STL.r <- Tbox$STL/Obox$STL kclu <- kclustering(X) is.kclustering(kclu)
Reports whether x is a 'MDSmap' object
is.MDSmap(x)is.MDSmap(x)
x |
an object to test. |
Returns TRUE if its argument is of class MDSmap and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- subset(Pbox, MIN >= 1500) data <- data.frame(data$PTS, data$P3M, data$P2M, data$OREB + data$DREB, data$AST, data$TOV,data$STL, data$BLK) mds <- MDSmap(data) is.MDSmap(mds)data <- subset(Pbox, MIN >= 1500) data <- data.frame(data$PTS, data$P3M, data$P2M, data$OREB + data$DREB, data$AST, data$TOV,data$STL, data$BLK) mds <- MDSmap(data) is.MDSmap(mds)
Reports whether x is a 'shotperformance' object
is.shotperformance(x)is.shotperformance(x)
x |
an object to test. |
Returns TRUE if its argument is of class shotperformance and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP_data = PbP, team_name = "GSW", player_data=Pbox, team_data = Tadd) PbP <- shotclock(PbP_data = PbP, sec_14_after_oreb = FALSE, team_data = Tadd) shotperf <- shotperformance(PbP_data = PbP, player_data = Pbox, team_data = Tadd, shotclock_interval = c(0, 2) , shot_type = "2P" ) is.shotperformance(shotperf)PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP_data = PbP, team_name = "GSW", player_data=Pbox, team_data = Tadd) PbP <- shotclock(PbP_data = PbP, sec_14_after_oreb = FALSE, team_data = Tadd) shotperf <- shotperformance(PbP_data = PbP, player_data = Pbox, team_data = Tadd, shotclock_interval = c(0, 2) , shot_type = "2P" ) is.shotperformance(shotperf)
Reports whether x is a 'simplereg' object
is.simplereg(x)is.simplereg(x)
x |
an object to test. |
Returns TRUE if its argument is of class simplereg and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.sel <- subset(Pbox, MIN >= 500) X <- Pbox.sel$AST/Pbox.sel$MIN Y <- Pbox.sel$TOV/Pbox.sel$MIN Pl <- Pbox.sel$Player out <- simplereg(x=X, y=Y, type="lin") is.simplereg(out)Pbox.sel <- subset(Pbox, MIN >= 500) X <- Pbox.sel$AST/Pbox.sel$MIN Y <- Pbox.sel$TOV/Pbox.sel$MIN Pl <- Pbox.sel$Player out <- simplereg(x=X, y=Y, type="lin") is.simplereg(out)
Reports whether x is a 'variability' object
is.variability(x)is.variability(x)
x |
an object to test. |
Returns TRUE if its argument is of class variability and FALSE otherwise.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.BC <- subset(Pbox, Team=="Oklahoma City Thunder" & MIN >= 500, select=c("P2p","P3p","FTp","P2A","P3A","FTA")) out <- variability(data=Pbox.BC, data.var=c("P2p","P3p","FTp"), size.var=c("P2A","P3A","FTA"), weight=TRUE) is.variability(out)Pbox.BC <- subset(Pbox, Team=="Oklahoma City Thunder" & MIN >= 500, select=c("P2p","P3p","FTp","P2A","P3A","FTA")) out <- variability(data=Pbox.BC, data.var=c("P2p","P3p","FTp"), size.var=c("P2A","P3A","FTA"), weight=TRUE) is.variability(out)
K-means cluster analysis
kclustering( data, k = NULL, labels = NULL, nclumax = 10, nruns = 10, iter.max = 50, algorithm = "Hartigan-Wong" )kclustering( data, k = NULL, labels = NULL, nclumax = 10, nruns = 10, iter.max = 50, algorithm = "Hartigan-Wong" )
data |
numeric data frame. |
k |
integer, number of clusters. |
labels |
character, row labels. |
nclumax |
integer, maximum number of clusters (when |
nruns |
integer, run the k-means algorithm |
iter.max |
integer, maximum number of iterations allowed in k-means clustering (see kmeans). |
algorithm |
character, the algorithm used in k-means clustering (see kmeans). |
The kclustering function performs a preliminary standardization of columns in data.
A kclustering object.
If k is NULL, the kclustering object is a list of 3 elements:
k NULL
clusterRange integer vector, values of k (from 1 to nclumax) at which the variance between of the clusterization is evaluated
VarianceBetween numeric vector, values of the variance between evaluated for k in clusterRange
If k is not NULL, the kclustering object is a list of 4 elements:
k integer, number of clusters
Subjects data frame, subjects' cluster identifiers
ClusterList list, clusters' composition
Profiles data frame, clusters' profiles, i.e. the average of the variables within clusters and the cluster eterogeineity index (CHI)
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
FF <- fourfactors(Tbox,Obox) X <- with(FF, data.frame(OD.Rtg=ORtg/DRtg, F1.r=F1.Def/F1.Off, F2.r=F2.Off/F2.Def, F3.O=F3.Def, F3.D=F3.Off)) X$P3M <- Tbox$P3M X$STL.r <- Tbox$STL/Obox$STL kclu1 <- kclustering(X) plot(kclu1) kclu2 <- kclustering(X, k=9) plot(kclu2)FF <- fourfactors(Tbox,Obox) X <- with(FF, data.frame(OD.Rtg=ORtg/DRtg, F1.r=F1.Def/F1.Off, F2.r=F2.Off/F2.Def, F3.O=F3.Def, F3.D=F3.Off)) X$P3M <- Tbox$P3M X$STL.r <- Tbox$STL/Obox$STL kclu1 <- kclustering(X) plot(kclu1) kclu2 <- kclustering(X, k=9) plot(kclu2)
Multidimensional scaling (MDS) in 2 dimensions
MDSmap(data, std = TRUE)MDSmap(data, std = TRUE)
data |
a numeric matrix, data frame or |
std |
logical; if TRUE, |
If data is an object of class "dist", std is not active and data is directly inputted into MASS::isoMDS.
An object of class MDSmap, i.e. a list with 4 objects:
points, a 2-column vector of the fitted configuration (see isoMDS);
stress, the final stress achieved in percent (see isoMDS);
data, the input data frame;
std, the logical std input.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- with(Pbox, data.frame(PTS, P3M, P2M, REB=OREB+DREB, AST, TOV, STL, BLK)) selp <- which(Pbox$MIN >= 1500) data <- data[selp, ] id <- Pbox$Player[selp] mds <- MDSmap(data) plot(mds, labels=id, z.var="P2M", level.plot=FALSE, palette=rainbow)data <- with(Pbox, data.frame(PTS, P3M, P2M, REB=OREB+DREB, AST, TOV, STL, BLK)) selp <- which(Pbox$MIN >= 1500) data <- data[selp, ] id <- Pbox$Player[selp] mds <- MDSmap(data) plot(mds, labels=id, z.var="P2M", level.plot=FALSE, palette=rainbow)
In this data frame cases (rows) are teams and variables (columns) are referred to achievements of the opponents in the NBA 2017-2018 Championship
OboxObox
A data frame with 30 rows and 23 variables:
Analyzed team, character
Games Played, numeric
Minutes Played, numeric
Points Made, numeric
Games won, numeric
Games lost, numeric
2-Point Field Goals (Made), numeric
2-Point Field Goals (Attempted), numeric
2-Point Field Goals (Percentage), numeric
3-Point Field Goals (Made), numeric
3-Point Field Goals (Attempted), numeric
3-Point Field Goals (Percentage), numeric
Free Throws (Made), numeric
Free Throws (Attempted), numeric
Free Throws (Percentage), numeric
Offensive Rebounds, numeric
Defensive Rebounds, numeric
Assists, numeric
Turnovers, numeric
Steals, numeric
Blocks, numeric
Personal Fouls, numeric
Plus/Minus, numeric
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
In this data frame, cases (rows) are players and variables (columns) are referred to the individual achievements in the NBA 2017-2018 Championship
PboxPbox
A data.frame with 605 rows and 22 variables:
Analyzed team, character
Analyzed player, character
Games Played, numeric
Minutes Played, numeric
Points Made, numeric
2-Point Field Goals (Made), numeric
2-Point Field Goals (Attempted), numeric
2-Point Field Goals (Percentage), numeric
3-Point Field Goals (Made), numeric
3-Point Field Goals (Attempted), numeric
3-Point Field Goals (Percentage), numeric
Free Throws (Made), numeric
Free Throws (Attempted), numeric
Free Throws (Percentage), numeric
Offensive Rebounds, numeric
Defensive Rebounds, numeric
Assists, numeric
Turnovers, numeric
Steals, numeric
Blocks, numeric
Personal Fouls, numeric
Plus/Minus, numeric
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
In this play-by-play data frame (NBA 2017-2018 Championship), the cases (rows) are the events occurred during the analyzed games and the variables (columns) are descriptions of the events in terms of type, time, players involved, score, area of the court.
PbP.BDBPbP.BDB
A data.frame with 37430 rows and 48 variables:
Identification code for the game
Season: years and type (Regular or Playoffs)
Date of the game
Five players on the court (away team; home team)
Quarter (>= 5: over-time)
Score of the away/home team
Time left in the quarter (h:mm:ss)
Time played in the quarter (h:mm:ss)
Time since the immediately preceding event (h:mm:ss)
Identification code for the play
Team responsible for the event
Type of event
Player who made the assist
Players for the jump ball
Player who blocked the shot
Player who entered/left the court
Sequence number of the free throw
Player who made the foul
Number of free throws accorded
Player responsible for the event
Scored points
Player who the jump ball is tipped to
Reason of the turnover
Result of the shot (made or missed)
Player who stole the ball
Type of play
Field shots: distance from the basket
Coordinates of the shooting player. original: tracking coordinate system half court, (0,0) center of the basket; converted: coordinates in feet full court, (0,0) bottom-left corner
Textual description of the event
This data set has been kindly made available by BigDataBall, a data provider which leverages computer-vision technologies to richen and extend sports datasets with lots of unique metrics. Since its establishment, BigDataBall has also supported many academic studies and is referred as a reliable source of validated and verified stats for NBA, MLB, NFL and WNBA.
The functions of BasketballAnalyzeR requiring play-by-play data as input need a data frame with some additional variables with respect to PbP.BDB. It can be obtained by means of the function PbPmanipulation.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
https://github.com/sndmrc/BasketballAnalyzeR
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Adapts the standard file supplied by BigDataBall to the format required by BasketballAnalyzeR
PbPmanipulation(data, period.length = 12, overtime.length = 5)PbPmanipulation(data, period.length = 12, overtime.length = 5)
data |
a play-by-play data frame supplied by BigDataBall. |
period.length |
numeric, the length of a quarter in minutes (default: 12 minutes as in NBA) |
overtime.length |
numeric, the length of an overtime period in minutes (default: 5 minutes as in NBA) |
A play-by-play data frame.
The data frame generated by PbPmanipulation has the same variables of PbP.BDB (when necessary, coerced from one data type to another, e.g from factor to numeric) plus the following five additional variables:
periodTime, time played in the quarter (in seconds)
totalTime, time played in the match (in seconds)
playlength, time since the immediately preceding event (in seconds)
ShotType, type of shot (FT, 2P, 3P)
oppTeam, name of the opponent team
hometeam, name of the home team (generated conditionally on the presence of the variable home_score)
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB)PbP <- PbPmanipulation(PbP.BDB)
Plots a network from a 'assistnet' object
## S3 method for class 'assistnet' plot( x, layout = "kamadakawai", layout.par = list(), edge.thr = 0, edge.col.lim = NULL, edge.col.lab = NULL, node.size = NULL, node.size.lab = NULL, node.col = NULL, node.col.lim = NULL, node.col.lab = NULL, node.pal = colorRampPalette(c("white", "blue", "red")), edge.pal = colorRampPalette(c("white", "blue", "red")), ... )## S3 method for class 'assistnet' plot( x, layout = "kamadakawai", layout.par = list(), edge.thr = 0, edge.col.lim = NULL, edge.col.lab = NULL, node.size = NULL, node.size.lab = NULL, node.col = NULL, node.col.lim = NULL, node.col.lab = NULL, node.pal = colorRampPalette(c("white", "blue", "red")), edge.pal = colorRampPalette(c("white", "blue", "red")), ... )
x |
an object of class |
layout |
character, network vertex layout algorithm (see |
layout.par |
a list of parameters for the network vertex layout algorithm (see |
edge.thr |
numeric, threshold for edge values; values below the threshold are set to 0. |
edge.col.lim |
numeric vector of length two providing limits of the scale for edge color. |
edge.col.lab |
character, label for edge color legend. |
node.size |
character, indicating the name of the variable for node size (one of the columns of the |
node.size.lab |
character, label for node size legend. |
node.col |
character, indicating the name of the variable for node color (one of the columns of the |
node.col.lim |
numeric vector of length two providing limits of the scale for node color. |
node.col.lab |
character, label for node color legend. |
node.pal |
color palette for node colors. |
edge.pal |
color palette for edge colors. |
... |
other graphical parameters. |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & player!="") out <- assistnet(PbP.GSW) plot(out, layout="circle", edge.thr=30, node.col="FGM_ASTp", node.size="ASTPTS")PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & player!="") out <- assistnet(PbP.GSW) plot(out, layout="circle", edge.thr=30, node.col="FGM_ASTp", node.size="ASTPTS")
Plots the correlation matrix and the correlation network from a 'corranalysis' object
## S3 method for class 'corranalysis' plot(x, horizontal = TRUE, title = NULL, ...)## S3 method for class 'corranalysis' plot(x, horizontal = TRUE, title = NULL, ...)
x |
an object of class |
horizontal |
logical; if TRUE, the two plots are arranged horizontally. |
title |
character, plot title. |
... |
other graphical parameters |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- data.frame(Pbox$PTS,Pbox$P3M,Pbox$P2M, Pbox$OREB + Pbox$DREB,Pbox$AST, Pbox$TOV,Pbox$STL,Pbox$BLK)/Pbox$MIN names(data) <- c("PTS","P3M","P2M","REB","AST","TOV","STL","BLK") data <- subset(data, Pbox$MIN >= 500) out <- corranalysis(data, threshold=0.5) plot(out)data <- data.frame(Pbox$PTS,Pbox$P3M,Pbox$P2M, Pbox$OREB + Pbox$DREB,Pbox$AST, Pbox$TOV,Pbox$STL,Pbox$BLK)/Pbox$MIN names(data) <- c("PTS","P3M","P2M","REB","AST","TOV","STL","BLK") data <- subset(data, Pbox$MIN >= 500) out <- corranalysis(data, threshold=0.5) plot(out)
Plot possessions, pace, offensive and defensive rating, and Four Factors from a 'fourfactors' object
## S3 method for class 'fourfactors' plot(x, title = NULL, ...)## S3 method for class 'fourfactors' plot(x, title = NULL, ...)
x |
an object of class |
title |
character, plot title. |
... |
other graphical parameters. |
The height of the bars in the two four factor plots are given by the difference between the team value and the average on the analyzed teams.
A list of four ggplot2 plots.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
selTeams <- c(2,6,10,11) FF <- fourfactors(Tbox[selTeams,], Obox[selTeams,]) plot(FF)selTeams <- c(2,6,10,11) FF <- fourfactors(Tbox[selTeams,], Obox[selTeams,]) plot(FF)
Plots hierarchical clustering from a 'hclustering' object
## S3 method for class 'hclustering' plot( x, title = NULL, profiles = FALSE, ncol.arrange = NULL, circlize = FALSE, horiz = TRUE, cex.labels = 0.7, colored.labels = TRUE, colored.branches = FALSE, rect = FALSE, lower.rect = NULL, min.mid.max = NULL, ... )## S3 method for class 'hclustering' plot( x, title = NULL, profiles = FALSE, ncol.arrange = NULL, circlize = FALSE, horiz = TRUE, cex.labels = 0.7, colored.labels = TRUE, colored.branches = FALSE, rect = FALSE, lower.rect = NULL, min.mid.max = NULL, ... )
x |
an object of class |
title |
character or vector of characters (when plotting radial plots of cluster profiles; see Value), plot title(s). |
profiles |
logical; if |
ncol.arrange |
integer, number of columns when arranging multiple grobs on a page (active when plotting radial plots of cluster profiles; see Value). |
circlize |
logical; if |
horiz |
logical; if |
cex.labels |
numeric, the magnification to be used for labels (active when plotting a dendrogram; see Value). |
colored.labels |
logical; if |
colored.branches |
logical; if |
rect |
logical; if |
lower.rect |
numeric, a value of how low should the lower part of the rect be (active when plotting a dendrogram; see option |
min.mid.max |
numeric vector with 3 elements: lower bound, middle dashed line, upper bound for radial axis (active when plotting radial plots of cluster profiles; see Value). |
... |
other graphical parameters. |
If x$k is NULL, plot.hclustering returns a single ggplot2 object, displaying the pattern of the explained variance vs the number of clusters.
If x$k is not NULL and profiles=FALSE, plot.hclustering returns a single ggplot2 object, displaying the dendrogram.
If x$k is not NULL and profiles=TRUE, plot.hclustering returns a list of ggplot2 objects, displaying the radial plots of the cluster profiles.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- with(Pbox, data.frame(PTS, P3M, REB=OREB+DREB, AST, TOV, STL, BLK, PF)) data <- subset(data, Pbox$MIN >= 1500) ID <- Pbox$Player[Pbox$MIN >= 1500] hclu1 <- hclustering(data) plot(hclu1) hclu2 <- hclustering(data, labels=ID, k=7) plot(hclu2)data <- with(Pbox, data.frame(PTS, P3M, REB=OREB+DREB, AST, TOV, STL, BLK, PF)) data <- subset(data, Pbox$MIN >= 1500) ID <- Pbox$Player[Pbox$MIN >= 1500] hclu1 <- hclustering(data) plot(hclu1) hclu2 <- hclustering(data, labels=ID, k=7) plot(hclu2)
Plot Lorenz curve from a 'inequality' object
## S3 method for class 'inequality' plot(x, title = NULL, ...)## S3 method for class 'inequality' plot(x, title = NULL, ...)
x |
an object of class |
title |
character, plot title. |
... |
other graphical parameters. |
A ggplot2 object.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.BN <- subset(Pbox, Team=="Brooklyn Nets") out <- inequality(Pbox.BN$PTS, nplayers=8) print(out) plot(out)Pbox.BN <- subset(Pbox, Team=="Brooklyn Nets") out <- inequality(Pbox.BN$PTS, nplayers=8) print(out) plot(out)
Plot k-means clustering from a 'kclustering' object
## S3 method for class 'kclustering' plot( x, title = NULL, ncol.arrange = NULL, min.mid.max = NULL, label.size = 2.5, ... )## S3 method for class 'kclustering' plot( x, title = NULL, ncol.arrange = NULL, min.mid.max = NULL, label.size = 2.5, ... )
x |
an object of class |
title |
character or vector of characters (when plotting radial plots of cluster profiles; see Value), plot title(s). |
ncol.arrange |
integer, number of columns when arranging multiple grobs on a page (active when plotting radial plots of cluster profiles; see Value). |
min.mid.max |
numeric vector with 3 elements: lower bound, middle dashed line, upper bound for radial axis (active when plotting radial plots of cluster profiles; see Value). |
label.size |
numeric; label font size (default 2.5). |
... |
other graphical parameters. |
If x$k is NULL, plot.kclustering returns a single ggplot2 object, displaying the pattern of the explained variance vs the number of clusters.
If x$k is not NULL, plot.kclustering returns a list of ggplot2 objects, displaying the radial plots of the cluster profiles.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
FF <- fourfactors(Tbox,Obox) X <- with(FF, data.frame(OD.Rtg=ORtg/DRtg, F1.r=F1.Def/F1.Off, F2.r=F2.Off/F2.Def, F3.O=F3.Def, F3.D=F3.Off)) X$P3M <- Tbox$P3M X$STL.r <- Tbox$STL/Obox$STL kclu1 <- kclustering(X) plot(kclu1) kclu2 <- kclustering(X, k=9) plot(kclu2)FF <- fourfactors(Tbox,Obox) X <- with(FF, data.frame(OD.Rtg=ORtg/DRtg, F1.r=F1.Def/F1.Off, F2.r=F2.Off/F2.Def, F3.O=F3.Def, F3.D=F3.Off)) X$P3M <- Tbox$P3M X$STL.r <- Tbox$STL/Obox$STL kclu1 <- kclustering(X) plot(kclu1) kclu2 <- kclustering(X, k=9) plot(kclu2)
Draws two-dimensional plots for multidimensional scaling (MDS) from a 'MDSmap' object
## S3 method for class 'MDSmap' plot( x, z.var = NULL, level.plot = TRUE, title = NULL, labels = NULL, repel_labels = FALSE, text_label = TRUE, label_size = 3, subset = NULL, col.subset = "gray50", zoom = NULL, palette = NULL, contour = FALSE, ncol.arrange = NULL, ... )## S3 method for class 'MDSmap' plot( x, z.var = NULL, level.plot = TRUE, title = NULL, labels = NULL, repel_labels = FALSE, text_label = TRUE, label_size = 3, subset = NULL, col.subset = "gray50", zoom = NULL, palette = NULL, contour = FALSE, ncol.arrange = NULL, ... )
x |
an object of class |
z.var |
character vector; defines the set of variables (available in the |
level.plot |
logical; if TRUE, draws a level plot, otherwise draws a scatter plot (not active if |
title |
character, plot title. |
labels |
character vector, labels for (x, y) points (only for single scatter plot). |
repel_labels |
logical; if |
text_label |
logical; if |
label_size |
numeric; label font size (default |
subset |
logical vector, to select a subset of points to be highlighted. |
col.subset |
character, color for the subset of points. |
zoom |
numeric vector with 4 elements; |
palette |
color palette. |
contour |
logical; if |
ncol.arrange |
integer, number of columns when arranging multiple grobs on a page. |
... |
other graphical parameters. |
A single ggplot2 plot or a list of ggplot2 plots
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data <- data.frame(Pbox$PTS, Pbox$P3M, Pbox$P2M, Pbox$OREB + Pbox$DREB, Pbox$AST, Pbox$TOV,Pbox$STL, Pbox$BLK) names(data) <- c('PTS','P3M','P2M','REB','AST','TOV','STL','BLK') selp <- which(Pbox$MIN >= 1500) data <- data[selp,] id <- Pbox$Player[selp] mds <- MDSmap(data) plot(mds, labels=id, z.var="P2M", level.plot=FALSE, palette=rainbow)data <- data.frame(Pbox$PTS, Pbox$P3M, Pbox$P2M, Pbox$OREB + Pbox$DREB, Pbox$AST, Pbox$TOV,Pbox$STL, Pbox$BLK) names(data) <- c('PTS','P3M','P2M','REB','AST','TOV','STL','BLK') selp <- which(Pbox$MIN >= 1500) data <- data[selp,] id <- Pbox$Player[selp] mds <- MDSmap(data) plot(mds, labels=id, z.var="P2M", level.plot=FALSE, palette=rainbow)
Plots a bubbleplot representing the data contained in the dataframe produced by the function 'shotperformance'
## S3 method for class 'shotperformance' plot(x, title = "Shooting performance", ...)## S3 method for class 'shotperformance' plot(x, title = "Shooting performance", ...)
x |
an object of class |
title |
character, plot title. |
... |
other graphical parameters. |
A ggplot2 object
Andrea Fox
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
P. Zuccolotto, M. Manisera and M. Sandri (2018) Big data analytics for modeling scoring probability in basketball: The effect of shooting under high pressure conditions. International Journal of Sports Science & Coaching.
# Draw the plot for the performances on 2 point shots, when the high pressure situation is # the one regarding shots taken when \code{shotclock} is between 0 and 2 PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data = Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data = Tadd) players_perf <- shotperformance(PbP, shotclock_interval = c(0, 2), player_data=Pbox, team_data = Tadd, shot_type = "2P", teams = "GSW") plot(players_perf)# Draw the plot for the performances on 2 point shots, when the high pressure situation is # the one regarding shots taken when \code{shotclock} is between 0 and 2 PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data = Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data = Tadd) players_perf <- shotperformance(PbP, shotclock_interval = c(0, 2), player_data=Pbox, team_data = Tadd, shot_type = "2P", teams = "GSW") plot(players_perf)
Plot simple regression from a 'simplereg' object
## S3 method for class 'simplereg' plot( x, labels = NULL, subset = NULL, Lx = 0.01, Ux = 0.99, Ly = 0.01, Uy = 0.99, title = "Simple regression", xtitle = NULL, ytitle = NULL, repel = TRUE, ... )## S3 method for class 'simplereg' plot( x, labels = NULL, subset = NULL, Lx = 0.01, Ux = 0.99, Ly = 0.01, Uy = 0.99, title = "Simple regression", xtitle = NULL, ytitle = NULL, repel = TRUE, ... )
x |
an object of class |
labels |
character, labels for subjects. |
subset |
an optional vector specifying a subset of observations to be highlighted in the graph or |
Lx |
numeric; if |
Ux |
numeric; if |
Ly |
numeric; if |
Uy |
numeric; if |
title |
character, plot title. |
xtitle |
character, x-axis label. |
ytitle |
character, y-axis label. |
repel |
logical, if |
... |
other graphical parameters. |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.sel <- subset(Pbox, MIN >= 500) X <- Pbox.sel$AST/Pbox.sel$MIN Y <- Pbox.sel$TOV/Pbox.sel$MIN Pl <- Pbox.sel$Player mod <- simplereg(x=X, y=Y, type="lin") plot(mod)Pbox.sel <- subset(Pbox, MIN >= 500) X <- Pbox.sel$AST/Pbox.sel$MIN Y <- Pbox.sel$TOV/Pbox.sel$MIN Pl <- Pbox.sel$Player mod <- simplereg(x=X, y=Y, type="lin") plot(mod)
Plots a variability diagram from a 'variability' object
## S3 method for class 'variability' plot( x, title = "Variability diagram", ylim = NULL, ylab = NULL, size.lim = NULL, max.circle = 25, n.circle = 4, leg.brk = NULL, leg.pos = "right", leg.just = "left", leg.nrow = NULL, leg.title = NULL, leg.title.pos = "top", ... )## S3 method for class 'variability' plot( x, title = "Variability diagram", ylim = NULL, ylab = NULL, size.lim = NULL, max.circle = 25, n.circle = 4, leg.brk = NULL, leg.pos = "right", leg.just = "left", leg.nrow = NULL, leg.title = NULL, leg.title.pos = "top", ... )
x |
an aobject of class |
title |
character, plot title. |
ylim |
numeric vector of length two, y-axis limits. |
ylab |
character, y-axis label. |
size.lim |
numeric vector of length two, set limits of the bubbles' size scale (see |
max.circle |
numeric, maximum size of the |
n.circle |
integer; if |
leg.brk |
numeric vector, breaks for bubbles' size legend (see |
leg.pos |
character or numeric vector of length two, legend position; available options |
leg.just |
character or numeric vector of length two; anchor point for positioning legend inside plot ( |
leg.nrow |
integer, number of rows of the bubbles' size legend. |
leg.title |
character, title of the bubbles' size legend. |
leg.title.pos |
character, position of the legend title; available options: |
... |
other graphical parameters. |
A ggplot2 object
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.BC <- subset(Pbox, Team=="Oklahoma City Thunder" & MIN >= 500, select=c("P2p","P3p","FTp","P2A","P3A","FTA")) out <- variability(data=Pbox.BC, data.var=c("P2p","P3p","FTp"), size.var=c("P2A","P3A","FTA"), weight=TRUE) plot(out, leg.brk=c(10,25,50,100,500,1000), max.circle=30)Pbox.BC <- subset(Pbox, Team=="Oklahoma City Thunder" & MIN >= 500, select=c("P2p","P3p","FTp","P2A","P3A","FTA")) out <- variability(data=Pbox.BC, data.var=c("P2p","P3p","FTp"), size.var=c("P2A","P3A","FTA"), weight=TRUE) plot(out, leg.brk=c(10,25,50,100,500,1000), max.circle=30)
Draws radial plots for player profiles
radialprofile( data, perc = FALSE, std = TRUE, title = NULL, ncol.arrange = NULL, min.mid.max = NULL, label.size = 2.5 )radialprofile( data, perc = FALSE, std = TRUE, title = NULL, ncol.arrange = NULL, min.mid.max = NULL, label.size = 2.5 )
data |
a data frame. |
perc |
logical; if |
std |
logical; if |
title |
character vector, titles for radial plots. |
ncol.arrange |
integer, number of columns in the grid of arranged plots. |
min.mid.max |
numeric vector with 3 elements: lower bound, middle dashed line, upper bound for radial axis. |
label.size |
numeric; label font size (default 2.5). |
A list of ggplot2 radial plots or, if ncol.arrange=NULL, a single ggplot2 plot of arranged radial plots
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
data("Pbox") Pbox.PG <- Pbox[1:6,] X <- data.frame(Pbox.PG$P2M, Pbox.PG$P3M, Pbox.PG$OREB+Pbox.PG$DREB, Pbox.PG$AST, Pbox.PG$TO)/Pbox.PG$MIN names(X) <- c("P2M","P3M","REB","AST","TO") radialprofile(data=X, ncol.arrange=3, title=Pbox.PG$Player)data("Pbox") Pbox.PG <- Pbox[1:6,] X <- data.frame(Pbox.PG$P2M, Pbox.PG$P3M, Pbox.PG$OREB+Pbox.PG$DREB, Pbox.PG$AST, Pbox.PG$TO)/Pbox.PG$MIN names(X) <- c("P2M","P3M","REB","AST","TO") radialprofile(data=X, ncol.arrange=3, title=Pbox.PG$Player)
Draws a scatter plot or a matrix of scatter plots
scatterplot( data, data.var, z.var = NULL, palette = NULL, labels = NULL, repel_labels = FALSE, text_label = TRUE, label_size = 3, subset = NULL, col.subset = "gray50", zoom = NULL, title = NULL, legend = TRUE, upper = list(continuous = "cor", combo = "box_no_facet", discrete = "facetbar", na = "na"), lower = list(continuous = "points", combo = "facethist", discrete = "facetbar", na = "na"), diag = list(continuous = "densityDiag", discrete = "barDiag", na = "naDiag") )scatterplot( data, data.var, z.var = NULL, palette = NULL, labels = NULL, repel_labels = FALSE, text_label = TRUE, label_size = 3, subset = NULL, col.subset = "gray50", zoom = NULL, title = NULL, legend = TRUE, upper = list(continuous = "cor", combo = "box_no_facet", discrete = "facetbar", na = "na"), lower = list(continuous = "points", combo = "facethist", discrete = "facetbar", na = "na"), diag = list(continuous = "densityDiag", discrete = "barDiag", na = "naDiag") )
data |
an object of class |
data.var |
character or numeric vector, name or column number of variables (in |
z.var |
character or number, name or column number of variable (in |
palette |
color palette (active when plotting a single scatter plot; see Value). |
labels |
character vector, labels for points (active when plotting a single scatter plot, see Value). |
repel_labels |
logical; if |
text_label |
logical; if |
label_size |
numeric; label font size (default |
subset |
logical or numeric vector, to select a subset of points to be highlighted (active when plotting a single scatter plot; see Value). |
col.subset |
character, color for the labels and rectangles of highlighted points (active when plotting a single scatter plot; see Value). |
zoom |
numeric vector with 4 elements; |
title |
character, plot title. |
legend |
logical, if |
upper |
list, may contain the variables |
lower |
list, may contain the variables |
diag |
list, may contain the variables |
If length(data.var)=2, the variable specified in z.var can be numeric or factor; if length(data.var)>2, the variable specified in z.var must be a factor.
A ggplot2 object with a single scatter plot if length(data.var)=2 or a matrix of scatter plots if length(data.var)>2.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
# Single scatter plot Pbox.sel <- subset(Pbox, MIN>= 500) X <- data.frame(AST=Pbox.sel$AST/Pbox.sel$MIN,TOV=Pbox.sel$TOV/Pbox.sel$MIN) X$PTSpm <- Pbox.sel$PTS/Pbox.sel$MIN mypal <- colorRampPalette(c("blue","yellow","red")) scatterplot(X, data.var=c("AST","TOV"), z.var="PTSpm", labels=1:nrow(X), palette=mypal) # Matrix of scatter plots data <- Pbox[1:50, c("PTS","P3M","P2M","OREB","Team")] scatterplot(data, data.var=1:4, z.var="Team")# Single scatter plot Pbox.sel <- subset(Pbox, MIN>= 500) X <- data.frame(AST=Pbox.sel$AST/Pbox.sel$MIN,TOV=Pbox.sel$TOV/Pbox.sel$MIN) X$PTSpm <- Pbox.sel$PTS/Pbox.sel$MIN mypal <- colorRampPalette(c("blue","yellow","red")) scatterplot(X, data.var=c("AST","TOV"), z.var="PTSpm", labels=1:nrow(X), palette=mypal) # Matrix of scatter plots data <- Pbox[1:50, c("PTS","P3M","P2M","OREB","Team")] scatterplot(data, data.var=1:4, z.var="Team")
Computes the score difference between the two teams in the match
scoredifference(PbP_data, team_name, player_data, team_data)scoredifference(PbP_data, team_name, player_data, team_data)
PbP_data |
a play-by-play data frame, previously handled by |
team_name |
name of the team we are interested in. The name can be either shortened (e.g. CLE) or extended (e.g. Cleveland Cavaliers) |
player_data |
dataframe containing the boxscore data of all players of a particula season. We need it to know the players who have played at least one match for a team during the season. This dataframe might be substituted by a dataframe which has a column |
team_data |
dataframe, contains several data regarding the teams in the NBA. Inside this function it is used only to check if |
The score difference computed by the function can be different from the simple difference between the score of the home team and the one of the away team, as we have to take account of the points scored during an action. Indeed, the value of score.diff indicates the difference in the score while the action was played
the initial play-by-play dataframe, with two additional columns:
score.diff: difference between the score of team_name and the score of the opposite team (see details for more informations)
*isHome: boolean which indicates if team_name is the home team in that play-by-play row
Andrea Fox
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
P. Zuccolotto, M. Manisera and M. Sandri (2018) Big data analytics for modeling scoring probability in basketball: The effect of shooting under high pressure conditions. International Journal of Sports Science & Coaching.
PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name="GSW", player_data=Pbox, team_data=Tadd)PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name="GSW", player_data=Pbox, team_data=Tadd)
Plots scoring probability of shots as a function of a given variable
scoringprob( data, var, shot.type, players = NULL, bw = 20, period.length = 12, xlab = NULL, x.range = "auto", title = NULL, palette = gg_color_hue, team = TRUE, col.team = "dodgerblue", legend = TRUE )scoringprob( data, var, shot.type, players = NULL, bw = 20, period.length = 12, xlab = NULL, x.range = "auto", title = NULL, palette = gg_color_hue, team = TRUE, col.team = "dodgerblue", legend = TRUE )
data |
a data frame whose rows are shots and with the following columns: |
var |
character, the string giving the name of the numerical variable according to which the scoring probability is estimated. Available options: |
shot.type |
character, the type of shots to be analyzed; available options: |
players |
subset of players to be displayed (optional; it can be used only if the |
bw |
numeric, the smoothing bandwidth of the kernel density estimator (see ksmooth). |
period.length |
numeric, the length of a quarter in minutes (default: 12 minutes as in NBA). |
xlab |
character, x-axis label. |
x.range |
numerical vector or character; available options: |
title |
character, plot title. |
palette |
color palette. |
team |
character; if |
col.team |
character, color of the scoring probability line for all the shots in data. |
legend |
character; if |
The data data frame could also be a play-by-play dataset provided that rows corresponding to events different from shots have NA in the ShotType variable.
Required columns:
result, a factor with the following levels: "made" for made shots, "miss" for missed shots, and "" for events different from shots
ShotType, a factor with the following levels: "2P", "3P", "FT" (and NA for events different from shots)
player, a factor with the name of the player who made the shot
playlength, a numeric variable with time between the shot and the immediately preceding event
periodTime, a numeric variable with seconds played in the quarter when the shot is attempted
totalTime, a numeric variable with seconds played in the whole match when the shot is attempted
shot_distance, a numeric variable with the distance of the shooting player from the basket (in feet)
A ggplot2 plot
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & result!="") players <- c("Kevin Durant","Draymond Green","Klay Thompson") scoringprob(data=PbP.GSW, shot.type="2P", players=players, var="shot_distance", col.team="gray")PbP <- PbPmanipulation(PbP.BDB) PbP.GSW <- subset(PbP, team=="GSW" & result!="") players <- c("Kevin Durant","Draymond Green","Klay Thompson") scoringprob(data=PbP.GSW, shot.type="2P", players=players, var="shot_distance", col.team="gray")
Computes the probability of scoring certain shot types in certain conditions, by looking at the result of the shots in the PbP provided
scoringprobability( PbP_data, team_name = "", shotclock_interval = c(0, 24), totaltime = 0, score_difference = c(-100, 100), shot_type = "field", team_data )scoringprobability( PbP_data, team_name = "", shotclock_interval = c(0, 24), totaltime = 0, score_difference = c(-100, 100), shot_type = "field", team_data )
PbP_data |
a play-by-play dataframe, previously handled by the PbPmanipulation function |
team_name |
character, if the play-by-play dataframe given as an input contains data for multiple teams, this parameters filters only the shots of the team we are interested in |
shotclock_interval |
vector of two numeric values or single numeric value, condition on the value of shotclock of the shots that will be considered |
totaltime |
numeric value, condition on the value of totalTime of the shots that will be considered |
score_difference |
vector of two numeric values or single numeric value, condition on the value of shotclock of the shots that will be considered |
shot_type |
character, the type of shots to be analyzed; available options: "2P", "3P", "FT", "field" |
team_data |
dataframe, contains several data regarding the teams in the NBA. Inside this function it is used only to check if |
numeric value, indicating the probability that a shots which respects all the conditions defined is made
Andrea Fox
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
P. Zuccolotto, M. Manisera and M. Sandri (2018) Big data analytics for modeling scoring probability in basketball: The effect of shooting under high pressure conditions. International Journal of Sports Science & Coaching.
# probability that a 2 point shot attempted by the Golden State Warriors # in the last two seconds of an action is made PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data=Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data=Tadd) scoringprobability(PbP, team_name = "GSW", shotclock_interval = c(0, 2), shot_type = "2P", team_data=Tadd) # probability that a 3 point shot attempted when the score difference is # between -5 and 1 is made PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data=Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data=Tadd) scoringprobability(PbP, team_name = "GSW", score_difference = c(-5, 1), shot_type = "3P", team_data=Tadd) # probability that a free throw attempted in the last 5 minutes is made PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data=Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data=Tadd) scoringprobability(PbP, team_name = "GSW", totaltime = 43, shot_type = "FT", team_data=Tadd)# probability that a 2 point shot attempted by the Golden State Warriors # in the last two seconds of an action is made PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data=Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data=Tadd) scoringprobability(PbP, team_name = "GSW", shotclock_interval = c(0, 2), shot_type = "2P", team_data=Tadd) # probability that a 3 point shot attempted when the score difference is # between -5 and 1 is made PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data=Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data=Tadd) scoringprobability(PbP, team_name = "GSW", score_difference = c(-5, 1), shot_type = "3P", team_data=Tadd) # probability that a free throw attempted in the last 5 minutes is made PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP, team_name = "GSW", player_data=Pbox, team_data=Tadd) PbP <- shotclock(PbP, sec_14_after_oreb = FALSE, team_data=Tadd) scoringprobability(PbP, team_name = "GSW", totaltime = 43, shot_type = "FT", team_data=Tadd)
Plots different kinds of charts based on shot coordinates
shotchart( data, x, y, z = NULL, z.fun = median, result = NULL, type = NULL, scatter = FALSE, num.sect = 7, n = 1000, col.limits = c(NA, NA), courtline.col = "black", bg.col = "white", sectline.col = "white", text.col = "white", legend = FALSE, drop.levels = TRUE, pt.col = "black", pt.alpha = 0.5, nbins = 25, palette = "mixed" )shotchart( data, x, y, z = NULL, z.fun = median, result = NULL, type = NULL, scatter = FALSE, num.sect = 7, n = 1000, col.limits = c(NA, NA), courtline.col = "black", bg.col = "white", sectline.col = "white", text.col = "white", legend = FALSE, drop.levels = TRUE, pt.col = "black", pt.alpha = 0.5, nbins = 25, palette = "mixed" )
data |
A data frame whose rows are field shots and columns are half-court shot coordinates x and y, and optionally additional variables to be specified in |
x |
character, indicating the variable name of the x coordinate. |
y |
character, indicating the variable name of the y coordinate. |
z |
character, indicating the name of the variable used to color the points (if |
z.fun |
function (active when |
result |
character (active when |
type |
character, indicating the plot type; available option are |
scatter |
logical, if TRUE a scatter plot of the shots is added to the plot. |
num.sect |
integer (active when |
n |
integer (active when |
col.limits |
numeric vector, (active when |
courtline.col |
color of court lines. |
bg.col |
background color. |
sectline.col |
color of sector lines (active when |
text.col |
color of text annotation within sectors (active when |
legend |
logical, if TRUE a legend for |
drop.levels |
logical, if TRUE unused levels of the |
pt.col |
color of points in the scatter plot. |
pt.alpha |
numeric, transparency of points in the scatter plot. |
nbins |
integer (active when |
palette |
color palette; available options |
The data dataframe could also be a play-by-play dataset provided that rows corresponding to events different from field shots have missing x and y coordinates.
x and y coordinates must be expressed in feets; the origin of the axes is positioned at the center of the field.
A ggplot2 object.
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
drawNBAcourt, geom_density_2d, geom_hex
PbP <- PbPmanipulation(PbP.BDB) subdata <- subset(PbP, player=="Kevin Durant") subdata$xx <- subdata$original_x/10 subdata$yy <- subdata$original_y/10-41.75 shotchart(data=subdata, x="xx", y="yy", scatter=TRUE) shotchart(data=subdata, x="xx", y="yy", scatter=TRUE, z="result") shotchart(data=subdata, x="xx", y="yy", scatter=TRUE, z="result", bg.col="black", courtline.col="white", palette="hot") shotchart(data=subdata, x="xx", y="yy", result="result", type="sectors", sectline.col="gray", text.col="red") shotchart(data=subdata, x="xx", y="yy", z="playlength", result="result", type="sectors", num.sect=5) shotchart(data=subdata, x="xx", y="yy", type="density-polygons", palette="bwr") shotchart(data=subdata, x="xx", y="yy", type="density-raster", scatter=TRUE, pt.col="tomato", pt.alpha=0.1) shotchart(data=subdata, x="xx", y="yy", type="density-hexbin", nbins=30)PbP <- PbPmanipulation(PbP.BDB) subdata <- subset(PbP, player=="Kevin Durant") subdata$xx <- subdata$original_x/10 subdata$yy <- subdata$original_y/10-41.75 shotchart(data=subdata, x="xx", y="yy", scatter=TRUE) shotchart(data=subdata, x="xx", y="yy", scatter=TRUE, z="result") shotchart(data=subdata, x="xx", y="yy", scatter=TRUE, z="result", bg.col="black", courtline.col="white", palette="hot") shotchart(data=subdata, x="xx", y="yy", result="result", type="sectors", sectline.col="gray", text.col="red") shotchart(data=subdata, x="xx", y="yy", z="playlength", result="result", type="sectors", num.sect=5) shotchart(data=subdata, x="xx", y="yy", type="density-polygons", palette="bwr") shotchart(data=subdata, x="xx", y="yy", type="density-raster", scatter=TRUE, pt.col="tomato", pt.alpha=0.1) shotchart(data=subdata, x="xx", y="yy", type="density-hexbin", nbins=30)
Computes, for each action, an estimate of the value of the shotclock when the action has ended
shotclock( PbP_data, team_data, sec_14_after_oreb = FALSE, report = FALSE, verbose = FALSE, seconds_added_after_made_shot = 2, max_error_threshold = 4 )shotclock( PbP_data, team_data, sec_14_after_oreb = FALSE, report = FALSE, verbose = FALSE, seconds_added_after_made_shot = 2, max_error_threshold = 4 )
PbP_data |
a play-by-play dataframe, previously handled by the function PbPmanipulation |
team_data |
dataframe, contains several data regarding the teams in the NBA. Inside this function it is used only to check if |
sec_14_after_oreb |
boolean, it indicates if the shotclock has been set to 14 seconds in certain situations. It has to be true if the data have been recorded after the 2018-19 season. The default value is |
report |
boolean, if TRUE, the function prints a few details about some data which have a negative value of shotclock (and therefore have been correceted) |
verbose |
boolean, if TRUE, adds some comments about the computations |
seconds_added_after_made_shot |
numeric value, after a shot is made the period clock is not stopped (unless it is in the last minutes of each quarter), hence a certain number of seconds has to be added in order to take account of the seconds taken for the inbound pass |
max_error_threshold |
numeric value, some errors still occur in the data and some negative values of shotclock are produced (in general due to some delay between the end of the action and its registration). This parameters indicates the maximum absolute value of negative shotclock which is arbitrarily fixed to a positive value; the values of shotclock below this threshold are set as NAs |
It is necessary that the name of the team is contained in the column corresponding to the description
The play-by-play data, with the additional data regarding the value of shotclock and the boolean indicating whether the action has started with a value of shotclock equal to 14 seconds
Andrea Fox
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
P. Zuccolotto, M. Manisera and M. Sandri (2018) Big data analytics for modeling scoring probability in basketball: The effect of shooting under high pressure conditions. International Journal of Sports Science & Coaching.
PbP <- PbPmanipulation(PbP.BDB) PbP <- shotclock(PbP_data = PbP, team_data = Tadd)PbP <- PbPmanipulation(PbP.BDB) PbP <- shotclock(PbP_data = PbP, team_data = Tadd)
Computes, for each player of a specific team, its performance measure
shotperformance( PbP_data, player_data, team_data, shotclock_interval = c(0, 24), totaltime = 0, score_difference = c(-100, 100), shot_type = "field", min_shots = 100, min_shots_high_pressure = 10, verbose = FALSE, teams = "all" )shotperformance( PbP_data, player_data, team_data, shotclock_interval = c(0, 24), totaltime = 0, score_difference = c(-100, 100), shot_type = "field", min_shots = 100, min_shots_high_pressure = 10, verbose = FALSE, teams = "all" )
PbP_data |
a play-by-play dataframe, previously handled by the functions PbPmanipulation, shotclock and scoredifference |
player_data |
dataframe containing the boxscore data of all players of a particula season. We need it to know the players who have played at least one match for a team during the season. This dataframe might be substituted by a dataframe which has a column |
team_data |
dataframe, contains several data regarding the teams in the NBA. Inside this function it is used only to check if |
shotclock_interval |
vector of two numeric values or single numeric value, condition on the value of shotclock of the shots that will be considered |
totaltime |
vector of two numeric values, condition on the value of score.diff of the shots that will be considered |
score_difference |
numeric value, condition on the value of totalTime of the shots that will be considered |
shot_type |
character, the type of shots to be analyzed; available options: "2P", "3P", "FT", "field" |
min_shots |
minimum value of total shots that a player must have attempted in order to qualify for the computation of the performance statistic |
min_shots_high_pressure |
minimum value of total shots that a player must have attempted in an high pressure situation in order to qualify for the computation of the performance statistic |
verbose |
boolean, if TRUE, adds some comments about the computations |
teams |
character or vector of characters, indicates the teams whose players we want to compute the performance statistics |
A dataframe containing, for each player which fulfils the conditions on the minimum number of shots, the value of the overall performance, the performance difference in S, the propensity to shoot in S, the total number of shots and the total number of shots in the high pressure situation defined
Andrea Fox
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
P. Zuccolotto, M. Manisera and M. Sandri (2018) Big data analytics for modeling scoring probability in basketball: The effect of shooting under high pressure conditions. International Journal of Sports Science & Coaching.
# We consider the high pressure situation of all shots attempted # when the shotclock value is below 2 seconds PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP_data = PbP, team_name = "GSW", player_data=Pbox, team_data = Tadd) PbP <- shotclock(PbP_data = PbP, sec_14_after_oreb = FALSE, team_data = Tadd) shotperformance(PbP_data = PbP, player_data = Pbox, team_data = Tadd, shotclock_interval = c(0, 2) , shot_type = "2P")# We consider the high pressure situation of all shots attempted # when the shotclock value is below 2 seconds PbP <- PbPmanipulation(PbP.BDB) PbP <- scoredifference(PbP_data = PbP, team_name = "GSW", player_data=Pbox, team_data = Tadd) PbP <- shotclock(PbP_data = PbP, sec_14_after_oreb = FALSE, team_data = Tadd) shotperformance(PbP_data = PbP, player_data = Pbox, team_data = Tadd, shotclock_interval = c(0, 2) , shot_type = "2P")
Simple linear and nonparametric regression
simplereg(x, y, type = "lin", sp = NULL)simplereg(x, y, type = "lin", sp = NULL)
x |
numerical vector, input x values. |
y |
numerical vector, input y values. |
type |
character, type of regression; available options are: |
sp |
numeric, parameter to control the degree of smoothing; span for local polynomial regression and bandwidth for ksmooth. |
An object of class simplereg, i.e. a list with the following objects:
Model, the output model (linear regression, local polynomial regression, or kernel smoothing)
R2, (in-sample) coefficient of determination
x, input x values
y, input y values
type, type of regression
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Pbox.sel <- subset(Pbox, MIN >= 500) X <- Pbox.sel$AST/Pbox.sel$MIN Y <- Pbox.sel$TOV/Pbox.sel$MIN Pl <- Pbox.sel$Player mod <- simplereg(x=X, y=Y, type="lin")Pbox.sel <- subset(Pbox, MIN >= 500) X <- Pbox.sel$AST/Pbox.sel$MIN Y <- Pbox.sel$TOV/Pbox.sel$MIN Pl <- Pbox.sel$Player mod <- simplereg(x=X, y=Y, type="lin")
In this data frame, the cases (rows) are the analyzed teams and the variables (columns) are qualitative information such as Conference, Division, final rank, qualification for Playoffs for the NBA 2017-2018 Championship.
TaddTadd
A data frame with 30 rows and 6 variables:
Analyzed team (long name), factor
Analyzed team (short name), factor
Conference, factor
Division, factor
Rank (end season), numeric
Playoff qualification (Yes or No), factor
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
In this data frame, cases (rows) are teams and variables (columns) are referred to team achievements in the different games in the NBA 2017-2018 Championship.
TboxTbox
A data frame with 30 rows and 23 variables:
Analyzed team, character
Games Played, numeric
Minutes Played, numeric
Points Made, numeric
Games won, numeric
Games lost, numeric
2-Point Field Goals (Made), numeric
2-Point Field Goals (Attempted), numeric
2-Point Field Goals (Percentage), numeric
3-Point Field Goals (Made), numeric
3-Point Field Goals (Attempted), numeric
3-Point Field Goals (Percentage), numeric
Free Throws (Made), numeric
Free Throws (Attempted), numeric
Free Throws (Percentage), numeric
Offensive Rebounds, numeric
Defensive Rebounds, numeric
Assists, numeric
Turnovers, numeric
Steals, numeric
Blocks, numeric
Personal Fouls, numeric
Plus/Minus, numeric
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
Calculate Tbox, Obox and Pbox
TOPboxes(data, team)TOPboxes(data, team)
data |
a play-by-play data frame |
team |
character, team |
A list with the following elements
Tbox, (completare descrizione)
Obox, (completare descrizione)
Pbox, (completare descrizione)
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
P. Zuccolotto and M. Manisera (2020) Basketball Data Science: With Applications in R. CRC Press.
library(operators) library(dplyr) PbP <- PbPmanipulation(PbP.BDB) PbP <- PbP %>% mutate(oreb = type %~% "rebound offensive", dreb = type %~% "rebound defensive", turnover = event_type=="turnover", PF = (event_type == "foul") & !(type %~% "technical") ) %>% mutate(across(c(player, assist, steal, block, h1:h5, a1:a5), as.character)) %>% as.data.frame() out <- TOPboxes(PbP, team="GSW")library(operators) library(dplyr) PbP <- PbPmanipulation(PbP.BDB) PbP <- PbP %>% mutate(oreb = type %~% "rebound offensive", dreb = type %~% "rebound defensive", turnover = event_type=="turnover", PF = (event_type == "foul") & !(type %~% "technical") ) %>% mutate(across(c(player, assist, steal, block, h1:h5, a1:a5), as.character)) %>% as.data.frame() out <- TOPboxes(PbP, team="GSW")
Variability analysis
variability(data, data.var, size.var, VC = TRUE, weight = FALSE)variability(data, data.var, size.var, VC = TRUE, weight = FALSE)
data |
a data frame. |
data.var |
a vector of variable names or of column numbers defining (numeric) variables whose variability will be analyzed by |
size.var |
a vector of variable names or of column numbers defining variables for weights (active only if |
VC |
logical; if |
weight |
logical; if TRUE, calculates weighted variation coefficients and standard deviations. |
A list with the following elements: ranges, standard deviations, variation coefficients, and two dataframes (data, size).
Marco Sandri, Paola Zuccolotto, Marica Manisera ([email protected])
Pbox.BC <- subset(Pbox, Team=="Oklahoma City Thunder" & MIN >= 500, select=c("P2p","P3p","FTp","P2A","P3A","FTA")) list_variability <- variability(data=Pbox.BC, data.var=c("P2p","P3p","FTp"), size.var=c("P2A","P3A","FTA"), weight=TRUE) print(list_variability) plot(list_variability, leg.brk=c(10,25,50,100,500,1000), max.circle=30)Pbox.BC <- subset(Pbox, Team=="Oklahoma City Thunder" & MIN >= 500, select=c("P2p","P3p","FTp","P2A","P3A","FTA")) list_variability <- variability(data=Pbox.BC, data.var=c("P2p","P3p","FTp"), size.var=c("P2A","P3A","FTA"), weight=TRUE) print(list_variability) plot(list_variability, leg.brk=c(10,25,50,100,500,1000), max.circle=30)