# Pour les explications sur le Quasi-experimental Plus/Minus, voir: http://vincentberthet.com/introducing-quasi-experimental-plusminus-qepm/#TP: Target Player
#CP: Compared Player
#T: Team (of TP or CP)
#O: team Opponent
#l: lineup# The raw data file can be downloaded on BasketballValue. Go to: "Downloads", "2007-2008" (for instance), and download the folder "List of each matchup of one unit against another". In the folder, the file is "matchups20072008reg20081211.text". Open the file and save it in csv format.
rm(list=ls())
d <- read.csv2 (file.choose())
head(d)## GameID StartTime EndTime ElapsedTime ElapsedSecs HomePlayer1ID
## 1 20071030HOULAL 00:00:01 00:00:00 00:00:01 1 189
## 2 20071030HOULAL 00:00:02 00:00:01 00:00:01 1 189
## 3 20071030HOULAL 00:01:32 00:00:02 00:01:30 90 189
## 4 20071030HOULAL 00:01:36 00:01:32 00:00:04 4 189
## 5 20071030HOULAL 00:04:30 00:01:36 00:02:54 174 189
## 6 20071030HOULAL 00:04:43 00:04:30 00:00:13 13 186
## HomePlayer2ID HomePlayer3ID HomePlayer4ID HomePlayer5ID AwayPlayer1ID
## 1 186 103 516 459 308
## 2 186 575 103 459 308
## 3 186 575 103 459 308
## 4 186 103 664 459 308
## 5 186 103 516 459 308
## 6 194 103 516 459 308
## AwayPlayer2ID AwayPlayer3ID AwayPlayer4ID AwayPlayer5ID HomePlayer1Name
## 1 318 305 302 62 Brown
## 2 198 508 318 305 Brown
## 3 198 318 305 302 Brown
## 4 198 318 305 302 Brown
## 5 198 508 305 302 Brown
## 6 198 508 305 302 Bryant
## HomePlayer2Name HomePlayer3Name HomePlayer4Name HomePlayer5Name
## 1 Bryant Fisher Turiaf Walton
## 2 Bryant Farmar Fisher Walton
## 3 Bryant Farmar Fisher Walton
## 4 Bryant Fisher Karl Walton
## 5 Bryant Fisher Turiaf Walton
## 6 Bynum Fisher Turiaf Walton
## AwayPlayer1Name AwayPlayer2Name AwayPlayer3Name AwayPlayer4Name
## 1 Alston James McGrady Yao
## 2 Alston Battier Hayes James
## 3 Alston Battier James McGrady
## 4 Alston Battier James McGrady
## 5 Alston Battier Hayes McGrady
## 6 Alston Battier Hayes McGrady
## AwayPlayer5Name StartScoreHome StartScoreAway EndScoreHome EndScoreAway
## 1 Wells 93 95 93 95
## 2 McGrady 92 95 93 95
## 3 Yao 83 92 92 95
## 4 Yao 80 92 83 92
## 5 Yao 76 82 80 92
## 6 Yao 76 82 76 82
## PointsScoredHome PointsScoredAway PlusMinusHome PlusMinusAway
## 1 0 0 0 0
## 2 1 0 1 -1
## 3 9 3 6 -6
## 4 3 0 3 -3
## 5 4 10 -6 6
## 6 0 0 0 0
## PossessionsHome PossessionsAway OffensiveRtgHome OffensiveRtgAway
## 1 0 0 NULL NULL
## 2 0 0 NULL NULL
## 3 4 5 225 60
## 4 1 0 300 NULL
## 5 5 6 80 166.667
## 6 0 0 NULL NULL
## OverallRtgHomevsAway OverallRtgAwayvsHome OffensiveReboundsHome
## 1 NULL NULL 0
## 2 NULL NULL 0
## 3 165 -165 0
## 4 NULL NULL 0
## 5 -86.6667 86.6667 0
## 6 NULL NULL 0
## OffensiveReboundsAway DefensiveReboundsHome DefensiveReboundsAway
## 1 0 0 0
## 2 0 0 0
## 3 0 1 0
## 4 0 0 0
## 5 0 1 3
## 6 0 0 0
## OffensiveReboundingRateHome OffensiveReboundingRateAway
## 1 NULL NULL
## 2 NULL NULL
## 3 NULL 0
## 4 NULL NULL
## 5 0 0
## 6 NULL NULL
## DefensiveReboundingRateHome DefensiveReboundingRateAway
## 1 NULL NULL
## 2 NULL NULL
## 3 1 NULL
## 4 NULL NULL
## 5 1 1
## 6 NULL NULLnrow_d <- nrow(d)
nrow_d## [1] 35458for (i in 1:nrow_d) {
d$Date[i] <- substr(d$GameID[i], start = 1, stop = 8)
d$AwayTeam[i] <- substr(d$GameID[i], start = 9, stop = 11)
d$HomeTeam[i] <- substr(d$GameID[i], start = 12, stop = 14)
d$GameName[i] <- toString(c(d$Date[i], d$AwayTeam[i], d$HomeTeam[i]))
}
d <- d[,-which(names(d) %in% c("GameID"))]
head(d)## StartTime EndTime ElapsedTime ElapsedSecs HomePlayer1ID HomePlayer2ID
## 1 00:00:01 00:00:00 00:00:01 1 189 186
## 2 00:00:02 00:00:01 00:00:01 1 189 186
## 3 00:01:32 00:00:02 00:01:30 90 189 186
## 4 00:01:36 00:01:32 00:00:04 4 189 186
## 5 00:04:30 00:01:36 00:02:54 174 189 186
## 6 00:04:43 00:04:30 00:00:13 13 186 194
## HomePlayer3ID HomePlayer4ID HomePlayer5ID AwayPlayer1ID AwayPlayer2ID
## 1 103 516 459 308 318
## 2 575 103 459 308 198
## 3 575 103 459 308 198
## 4 103 664 459 308 198
## 5 103 516 459 308 198
## 6 103 516 459 308 198
## AwayPlayer3ID AwayPlayer4ID AwayPlayer5ID HomePlayer1Name
## 1 305 302 62 Brown
## 2 508 318 305 Brown
## 3 318 305 302 Brown
## 4 318 305 302 Brown
## 5 508 305 302 Brown
## 6 508 305 302 Bryant
## HomePlayer2Name HomePlayer3Name HomePlayer4Name HomePlayer5Name
## 1 Bryant Fisher Turiaf Walton
## 2 Bryant Farmar Fisher Walton
## 3 Bryant Farmar Fisher Walton
## 4 Bryant Fisher Karl Walton
## 5 Bryant Fisher Turiaf Walton
## 6 Bynum Fisher Turiaf Walton
## AwayPlayer1Name AwayPlayer2Name AwayPlayer3Name AwayPlayer4Name
## 1 Alston James McGrady Yao
## 2 Alston Battier Hayes James
## 3 Alston Battier James McGrady
## 4 Alston Battier James McGrady
## 5 Alston Battier Hayes McGrady
## 6 Alston Battier Hayes McGrady
## AwayPlayer5Name StartScoreHome StartScoreAway EndScoreHome EndScoreAway
## 1 Wells 93 95 93 95
## 2 McGrady 92 95 93 95
## 3 Yao 83 92 92 95
## 4 Yao 80 92 83 92
## 5 Yao 76 82 80 92
## 6 Yao 76 82 76 82
## PointsScoredHome PointsScoredAway PlusMinusHome PlusMinusAway
## 1 0 0 0 0
## 2 1 0 1 -1
## 3 9 3 6 -6
## 4 3 0 3 -3
## 5 4 10 -6 6
## 6 0 0 0 0
## PossessionsHome PossessionsAway OffensiveRtgHome OffensiveRtgAway
## 1 0 0 NULL NULL
## 2 0 0 NULL NULL
## 3 4 5 225 60
## 4 1 0 300 NULL
## 5 5 6 80 166.667
## 6 0 0 NULL NULL
## OverallRtgHomevsAway OverallRtgAwayvsHome OffensiveReboundsHome
## 1 NULL NULL 0
## 2 NULL NULL 0
## 3 165 -165 0
## 4 NULL NULL 0
## 5 -86.6667 86.6667 0
## 6 NULL NULL 0
## OffensiveReboundsAway DefensiveReboundsHome DefensiveReboundsAway
## 1 0 0 0
## 2 0 0 0
## 3 0 1 0
## 4 0 0 0
## 5 0 1 3
## 6 0 0 0
## OffensiveReboundingRateHome OffensiveReboundingRateAway
## 1 NULL NULL
## 2 NULL NULL
## 3 NULL 0
## 4 NULL NULL
## 5 0 0
## 6 NULL NULL
## DefensiveReboundingRateHome DefensiveReboundingRateAway Date
## 1 NULL NULL 20071030
## 2 NULL NULL 20071030
## 3 1 NULL 20071030
## 4 NULL NULL 20071030
## 5 1 1 20071030
## 6 NULL NULL 20071030
## AwayTeam HomeTeam GameName
## 1 HOU LAL 20071030, HOU, LAL
## 2 HOU LAL 20071030, HOU, LAL
## 3 HOU LAL 20071030, HOU, LAL
## 4 HOU LAL 20071030, HOU, LAL
## 5 HOU LAL 20071030, HOU, LAL
## 6 HOU LAL 20071030, HOU, LALOn numérote les games
d$GameID <- rep(c(0), nrow(d))
d$GameID[1] <- 1
head(d)## StartTime EndTime ElapsedTime ElapsedSecs HomePlayer1ID HomePlayer2ID
## 1 00:00:01 00:00:00 00:00:01 1 189 186
## 2 00:00:02 00:00:01 00:00:01 1 189 186
## 3 00:01:32 00:00:02 00:01:30 90 189 186
## 4 00:01:36 00:01:32 00:00:04 4 189 186
## 5 00:04:30 00:01:36 00:02:54 174 189 186
## 6 00:04:43 00:04:30 00:00:13 13 186 194
## HomePlayer3ID HomePlayer4ID HomePlayer5ID AwayPlayer1ID AwayPlayer2ID
## 1 103 516 459 308 318
## 2 575 103 459 308 198
## 3 575 103 459 308 198
## 4 103 664 459 308 198
## 5 103 516 459 308 198
## 6 103 516 459 308 198
## AwayPlayer3ID AwayPlayer4ID AwayPlayer5ID HomePlayer1Name
## 1 305 302 62 Brown
## 2 508 318 305 Brown
## 3 318 305 302 Brown
## 4 318 305 302 Brown
## 5 508 305 302 Brown
## 6 508 305 302 Bryant
## HomePlayer2Name HomePlayer3Name HomePlayer4Name HomePlayer5Name
## 1 Bryant Fisher Turiaf Walton
## 2 Bryant Farmar Fisher Walton
## 3 Bryant Farmar Fisher Walton
## 4 Bryant Fisher Karl Walton
## 5 Bryant Fisher Turiaf Walton
## 6 Bynum Fisher Turiaf Walton
## AwayPlayer1Name AwayPlayer2Name AwayPlayer3Name AwayPlayer4Name
## 1 Alston James McGrady Yao
## 2 Alston Battier Hayes James
## 3 Alston Battier James McGrady
## 4 Alston Battier James McGrady
## 5 Alston Battier Hayes McGrady
## 6 Alston Battier Hayes McGrady
## AwayPlayer5Name StartScoreHome StartScoreAway EndScoreHome EndScoreAway
## 1 Wells 93 95 93 95
## 2 McGrady 92 95 93 95
## 3 Yao 83 92 92 95
## 4 Yao 80 92 83 92
## 5 Yao 76 82 80 92
## 6 Yao 76 82 76 82
## PointsScoredHome PointsScoredAway PlusMinusHome PlusMinusAway
## 1 0 0 0 0
## 2 1 0 1 -1
## 3 9 3 6 -6
## 4 3 0 3 -3
## 5 4 10 -6 6
## 6 0 0 0 0
## PossessionsHome PossessionsAway OffensiveRtgHome OffensiveRtgAway
## 1 0 0 NULL NULL
## 2 0 0 NULL NULL
## 3 4 5 225 60
## 4 1 0 300 NULL
## 5 5 6 80 166.667
## 6 0 0 NULL NULL
## OverallRtgHomevsAway OverallRtgAwayvsHome OffensiveReboundsHome
## 1 NULL NULL 0
## 2 NULL NULL 0
## 3 165 -165 0
## 4 NULL NULL 0
## 5 -86.6667 86.6667 0
## 6 NULL NULL 0
## OffensiveReboundsAway DefensiveReboundsHome DefensiveReboundsAway
## 1 0 0 0
## 2 0 0 0
## 3 0 1 0
## 4 0 0 0
## 5 0 1 3
## 6 0 0 0
## OffensiveReboundingRateHome OffensiveReboundingRateAway
## 1 NULL NULL
## 2 NULL NULL
## 3 NULL 0
## 4 NULL NULL
## 5 0 0
## 6 NULL NULL
## DefensiveReboundingRateHome DefensiveReboundingRateAway Date
## 1 NULL NULL 20071030
## 2 NULL NULL 20071030
## 3 1 NULL 20071030
## 4 NULL NULL 20071030
## 5 1 1 20071030
## 6 NULL NULL 20071030
## AwayTeam HomeTeam GameName GameID
## 1 HOU LAL 20071030, HOU, LAL 1
## 2 HOU LAL 20071030, HOU, LAL 0
## 3 HOU LAL 20071030, HOU, LAL 0
## 4 HOU LAL 20071030, HOU, LAL 0
## 5 HOU LAL 20071030, HOU, LAL 0
## 6 HOU LAL 20071030, HOU, LAL 0for(i in 2:nrow_d)
if(d$GameName[i] == d$GameName[i-1]) {d$GameID[i] <- d$GameID[i-1]} else {d$GameID[i] <- d$GameID[i-1]+1}
head(d)## StartTime EndTime ElapsedTime ElapsedSecs HomePlayer1ID HomePlayer2ID
## 1 00:00:01 00:00:00 00:00:01 1 189 186
## 2 00:00:02 00:00:01 00:00:01 1 189 186
## 3 00:01:32 00:00:02 00:01:30 90 189 186
## 4 00:01:36 00:01:32 00:00:04 4 189 186
## 5 00:04:30 00:01:36 00:02:54 174 189 186
## 6 00:04:43 00:04:30 00:00:13 13 186 194
## HomePlayer3ID HomePlayer4ID HomePlayer5ID AwayPlayer1ID AwayPlayer2ID
## 1 103 516 459 308 318
## 2 575 103 459 308 198
## 3 575 103 459 308 198
## 4 103 664 459 308 198
## 5 103 516 459 308 198
## 6 103 516 459 308 198
## AwayPlayer3ID AwayPlayer4ID AwayPlayer5ID HomePlayer1Name
## 1 305 302 62 Brown
## 2 508 318 305 Brown
## 3 318 305 302 Brown
## 4 318 305 302 Brown
## 5 508 305 302 Brown
## 6 508 305 302 Bryant
## HomePlayer2Name HomePlayer3Name HomePlayer4Name HomePlayer5Name
## 1 Bryant Fisher Turiaf Walton
## 2 Bryant Farmar Fisher Walton
## 3 Bryant Farmar Fisher Walton
## 4 Bryant Fisher Karl Walton
## 5 Bryant Fisher Turiaf Walton
## 6 Bynum Fisher Turiaf Walton
## AwayPlayer1Name AwayPlayer2Name AwayPlayer3Name AwayPlayer4Name
## 1 Alston James McGrady Yao
## 2 Alston Battier Hayes James
## 3 Alston Battier James McGrady
## 4 Alston Battier James McGrady
## 5 Alston Battier Hayes McGrady
## 6 Alston Battier Hayes McGrady
## AwayPlayer5Name StartScoreHome StartScoreAway EndScoreHome EndScoreAway
## 1 Wells 93 95 93 95
## 2 McGrady 92 95 93 95
## 3 Yao 83 92 92 95
## 4 Yao 80 92 83 92
## 5 Yao 76 82 80 92
## 6 Yao 76 82 76 82
## PointsScoredHome PointsScoredAway PlusMinusHome PlusMinusAway
## 1 0 0 0 0
## 2 1 0 1 -1
## 3 9 3 6 -6
## 4 3 0 3 -3
## 5 4 10 -6 6
## 6 0 0 0 0
## PossessionsHome PossessionsAway OffensiveRtgHome OffensiveRtgAway
## 1 0 0 NULL NULL
## 2 0 0 NULL NULL
## 3 4 5 225 60
## 4 1 0 300 NULL
## 5 5 6 80 166.667
## 6 0 0 NULL NULL
## OverallRtgHomevsAway OverallRtgAwayvsHome OffensiveReboundsHome
## 1 NULL NULL 0
## 2 NULL NULL 0
## 3 165 -165 0
## 4 NULL NULL 0
## 5 -86.6667 86.6667 0
## 6 NULL NULL 0
## OffensiveReboundsAway DefensiveReboundsHome DefensiveReboundsAway
## 1 0 0 0
## 2 0 0 0
## 3 0 1 0
## 4 0 0 0
## 5 0 1 3
## 6 0 0 0
## OffensiveReboundingRateHome OffensiveReboundingRateAway
## 1 NULL NULL
## 2 NULL NULL
## 3 NULL 0
## 4 NULL NULL
## 5 0 0
## 6 NULL NULL
## DefensiveReboundingRateHome DefensiveReboundingRateAway Date
## 1 NULL NULL 20071030
## 2 NULL NULL 20071030
## 3 1 NULL 20071030
## 4 NULL NULL 20071030
## 5 1 1 20071030
## 6 NULL NULL 20071030
## AwayTeam HomeTeam GameName GameID
## 1 HOU LAL 20071030, HOU, LAL 1
## 2 HOU LAL 20071030, HOU, LAL 1
## 3 HOU LAL 20071030, HOU, LAL 1
## 4 HOU LAL 20071030, HOU, LAL 1
## 5 HOU LAL 20071030, HOU, LAL 1
## 6 HOU LAL 20071030, HOU, LAL 1On sauvegarde le fichier brut comportant ces quelques mises en forme
save(d, file = "d.RData")library(psych)
library(prettyR)##
## Attaching package: 'prettyR'
##
## The following objects are masked from 'package:psych':
##
## describe, skew############################## FUNCTION START #############################
qepm <- function(x){
load("d.RData")
####################ON SELECTIONNE LES GAMES DANS LESQUELS LE TP A JOUE
game_ID_d <- unique(d$GameID)
game_ID_d
d$selection <- rep(c(0), nrow(d))
d
for (i in 1:length(game_ID_d))
if(x %in% d[d$GameID == i,]$HomePlayer1ID == TRUE | x %in% d[d$GameID == i,]$HomePlayer2ID == TRUE | x %in% d[d$GameID == i,]$HomePlayer3ID == TRUE | x %in% d[d$GameID == i,]$HomePlayer4ID == TRUE | x %in% d[d$GameID == i,]$HomePlayer5ID == TRUE | x %in% d[d$GameID == i,]$AwayPlayer1ID == TRUE | x %in% d[d$GameID == i,]$AwayPlayer2ID == TRUE | x %in% d[d$GameID == i,]$AwayPlayer3ID == TRUE | x %in% d[d$GameID == i,]$AwayPlayer4ID == TRUE | x %in% d[d$GameID == i,]$AwayPlayer5ID == TRUE) {d[d$GameID == i,]$selection <- rep(c(1), length(d[d$GameID == i,]$GameID))} else {d[d$GameID == i,]$selection <- rep(c(0), length(d[d$GameID == i,]$GameID))}
d
##
d2 <- subset(d, d$selection == 1)
d2
###################################################################################
####################ON FAIT EN SORTE QUE: UN GAME = UN OBJET
game_ID <- unique(d2$GameID)
game_ID
length(game_ID) ## Games played by TP
data_raw = list()
for (i in 1:length(game_ID)){
data_raw[[i]] <- d2[d2$GameID == game_ID[i],]
}
####################MAIN LOOP
result <- list()
for (k in 1:length(game_ID)) {
#for (k in 1:1) {
result[[k]] <- list()
####################ON DETERMINE SI L'EQUIPE DU TP EST "Home" ou "Away"
TP_home_away <- ifelse(x %in% data_raw[[k]]$HomePlayer1ID == TRUE | x %in% data_raw[[k]]$HomePlayer2ID == TRUE | x %in% data_raw[[k]]$HomePlayer3ID == TRUE | x %in% data_raw[[k]]$HomePlayer4ID == TRUE | x %in% data_raw[[k]]$HomePlayer5ID == TRUE, "Home", "Away")
team_name <- if(TP_home_away == "Home") {as.character(unique(data_raw[[k]]$HomeTeam))} else {as.character(unique(data_raw[[k]]$AwayTeam))}
opp_name <- if(TP_home_away == "Home") {as.character(unique(data_raw[[k]]$AwayTeam))} else {as.character(unique(data_raw[[k]]$HomeTeam))}
date <- as.character(unique(data_raw[[k]]$Date))
gameID <- as.character(unique(data_raw[[k]]$GameID))
####################ON IDENTIFIE LES DIFFERENTES LINEUPS
##LINEUPS DE L'EQUIPE DU TP ET DU CP
nrow_data_raw <- nrow(data_raw[[k]])
for (i in 1:nrow_data_raw) {
if(TP_home_away == "Home") {data_raw[[k]]$T_l_h[i] <- as.numeric(paste(as.character(sort(c(data_raw[[k]]$HomePlayer1ID[i],data_raw[[k]]$HomePlayer2ID[i],data_raw[[k]]$HomePlayer3ID[i],data_raw[[k]]$HomePlayer4ID[i],data_raw[[k]]$HomePlayer5ID[i]))), collapse = ''))} else {data_raw[[k]]$T_l_h[i] <- as.numeric(paste(as.character(sort(c(data_raw[[k]]$AwayPlayer1ID[i],data_raw[[k]]$AwayPlayer2ID[i],data_raw[[k]]$AwayPlayer3ID[i],data_raw[[k]]$AwayPlayer4ID[i],data_raw[[k]]$AwayPlayer5ID[i]))), collapse = ''))}
##LINEUPS DE L'EQUIPE EN FACE
if(TP_home_away == "Away") {data_raw[[k]]$O_l_h[i] <- as.numeric(paste(as.character(sort(c(data_raw[[k]]$HomePlayer1ID[i],data_raw[[k]]$HomePlayer2ID[i],data_raw[[k]]$HomePlayer3ID[i],data_raw[[k]]$HomePlayer4ID[i],data_raw[[k]]$HomePlayer5ID[i]))), collapse = ''))} else {data_raw[[k]]$O_l_h[i] <- as.numeric(paste(as.character(sort(c(data_raw[[k]]$AwayPlayer1ID[i],data_raw[[k]]$AwayPlayer2ID[i],data_raw[[k]]$AwayPlayer3ID[i],data_raw[[k]]$AwayPlayer4ID[i],data_raw[[k]]$AwayPlayer5ID[i]))), collapse = ''))}
}
################ON NUMEROTE LES LINEUPS
T_lineups <- cbind(data.frame(data_raw[[k]]$T_l_h), class=as.numeric(as.factor(do.call(paste, data.frame(data_raw[[k]]$T_l_h)))))
O_lineups <- cbind(data.frame(data_raw[[k]]$O_l_h), class=as.numeric(as.factor(do.call(paste, data.frame(data_raw[[k]]$O_l_h)))))
O_lineups
data_raw[[k]]$T_l <- T_lineups$class
data_raw[[k]]$O_l <- O_lineups$class
################ON IDENTIFIE LES LINEUPS UNIQUES
T_l_unique_raw <- data_raw[[k]][!duplicated(data_raw[[k]]$T_l),]
T_l_unique_raw
O_l_unique_raw <- data_raw[[k]][!duplicated(data_raw[[k]]$O_l),]
O_l_unique_raw
T_l_unique <- T_l_unique_raw[order(T_l_unique_raw$T_l),]
T_l_unique
O_l_unique <- O_l_unique_raw[order(O_l_unique_raw$O_l),]
O_l_unique
for (i in 1:nrow_data_raw)
data_raw[[k]]$c1[i] <- toString(c(data_raw[[k]]$T_l[i],data_raw[[k]]$O_l[i]))
c1_id <- cbind(data.frame(data_raw[[k]]$c1), class=as.numeric(as.factor(do.call(paste, data.frame(data_raw[[k]]$c1)))))
c1_id
data_raw[[k]]$c2 <- c1_id$class
for (i in 1:nrow_data_raw) {
data_raw[[k]]$occ[i] <- ifelse(length(which(data_raw[[k]]$c2 %in% data_raw[[k]]$c2[i])) == 1, 1, which(which(data_raw[[k]]$c2 %in% data_raw[[k]]$c2[i]) %in% i))
data_raw[[k]]$c4[i] <- toString(c(data_raw[[k]]$T_l[i],data_raw[[k]]$O_l[i],data_raw[[k]]$occ[i]))
}
################ON GENERE TOUTES LES COMPARAISONS DEUX A DEUX
comp_all_raw <- data.frame(t(combn(data_raw[[k]]$c4, 2)))
comp_all_raw
comp_all_raw$X1 <- as.character(comp_all_raw$X1)
comp_all_raw$X2 <- as.character(comp_all_raw$X2)
nrow_comp_all_raw <- nrow(comp_all_raw)
nrow_comp_all_raw
for(i in 1:nrow_comp_all_raw) {
comp_all_raw$PA_T_l[i] <- (strsplit(comp_all_raw$X1, ", ")[[i]])[1]
comp_all_raw$PB_T_l[i] <- (strsplit(comp_all_raw$X2, ", ")[[i]])[1]
comp_all_raw$PA_O_l[i] <- (strsplit(comp_all_raw$X1, ", ")[[i]])[2]
comp_all_raw$PB_O_l[i] <- (strsplit(comp_all_raw$X2, ", ")[[i]])[2]
comp_all_raw$PA_T_l_O_l_occ[i] <- (strsplit(comp_all_raw$X1, ", ")[[i]])[3]
comp_all_raw$PB_T_l_O_l_occ[i] <- (strsplit(comp_all_raw$X2, ", ")[[i]])[3]
}
comp_all_raw
###########ON RETIRE LES COMPARAISONS NON PERTINENTES (i.e. comparaisons de lineups identiques: PA_T_l == PB_T_l)
comp_all_raw$h <- ifelse(comp_all_raw$PA_T_l == comp_all_raw$PB_T_l, 1, 0)
comp_all_raw
comp_all <- subset(comp_all_raw, comp_all_raw$h != 1)
comp_all
nrow_comp_all <- nrow(comp_all)
nrow_comp_all
comp_all$PA_T_l <- as.numeric(comp_all$PA_T_l)
comp_all$PB_T_l <- as.numeric(comp_all$PB_T_l)
comp_all$PA_O_l <- as.numeric(comp_all$PA_O_l)
comp_all$PB_O_l <- as.numeric(comp_all$PB_O_l)
################ON IDENTIFIE LES JOUEURS DES LINEUPS UNIQUES
T_l_unique_players <- if(TP_home_away == "Home") {T_l_unique[,c('HomePlayer1ID','HomePlayer2ID','HomePlayer3ID','HomePlayer4ID','HomePlayer5ID')]} else {T_l_unique[,c('AwayPlayer1ID','AwayPlayer2ID','AwayPlayer3ID','AwayPlayer4ID','AwayPlayer5ID')]}
T_l_unique_players
O_l_unique_players <- if(TP_home_away == "Away") {O_l_unique[,c('HomePlayer1ID','HomePlayer2ID','HomePlayer3ID','HomePlayer4ID','HomePlayer5ID')]} else {O_l_unique[,c('AwayPlayer1ID','AwayPlayer2ID','AwayPlayer3ID','AwayPlayer4ID','AwayPlayer5ID')]}
O_l_unique_players
T_l_unique_players_transp <- data.frame(t(T_l_unique_players))
T_l_unique_players_transp
O_l_unique_players_transp <- data.frame(t(O_l_unique_players))
O_l_unique_players_transp
for(i in 1:nrow_comp_all) {
comp_all$PA_T_l_players[i] <- toString(T_l_unique_players_transp[,comp_all$PA_T_l[i]])
comp_all$PB_T_l_players[i] <- toString(T_l_unique_players_transp[,comp_all$PB_T_l[i]])
comp_all$PA_O_l_players[i] <- toString(O_l_unique_players_transp[,comp_all$PA_O_l[i]])
comp_all$PB_O_l_players[i] <- toString(O_l_unique_players_transp[,comp_all$PB_O_l[i]])
}
comp_all
################ON IDENTIFIE LES JOUEURS QUI DIFFERENT ENTRE LES LINEUPS (DE L'EQUIPE DU TP)
##
outersect <- function(x, y) {
sort(c(setdiff(x, y),
setdiff(y, x)))
}
##
for(i in 1:nrow_comp_all) {
comp_all$T_players_differing[i] <- toString(outersect(T_l_unique_players_transp[,comp_all$PA_T_l[i]], T_l_unique_players_transp[,comp_all$PB_T_l[i]]))
comp_all$n_differing[i] <- length(strsplit(comp_all$T_players_differing[i], ", ")[[1]])
}
comp_all
################ON SELECTIONNE LES COMPARAISONS "1" (SEULS DEUX JOUEURS DIFFERENT ENTRE LES LINEUPS)
comp_2 <- subset(comp_all, comp_all$n_differing == 2)
comp_2
nrow_comp_2 <- nrow(comp_2)
nrow_comp_2
################ON IDENTIFIE LES DEUX JOUEURS DANS CHAQUE COMPARAISON
for(i in 1:nrow_comp_2) {
comp_2$player_A[i] <- (strsplit(comp_2$T_players_differing, ", ")[[i]])[1]
comp_2$player_B[i] <- (strsplit(comp_2$T_players_differing, ", ")[[i]])[2]
}
comp_2
comp_2$player_A <- as.numeric(comp_2$player_A)
comp_2$player_B <- as.numeric(comp_2$player_B)
################ON SELECTIONNE LES COMPARAISONS QUI IMPLIQUENT LE TP
comp_2$h2 <- ifelse(comp_2$player_A == x | comp_2$player_B == x, 1 ,0)
comp_2
comp_2_TP <- subset(comp_2, comp_2$h2 == 1)
comp_2_TP
nrow_comp_2_TP <- nrow(comp_2_TP)
nrow_comp_2_TP
if(nrow_comp_2_TP == 0) {result[[k]] <- 0} else {
################ON IDENTIFIE LE CP DANS CHAQUE COMPARAISON
comp_2_TP$TP_ID <- rep(c(x), nrow_comp_2_TP)
comp_2_TP$CP_ID <- ifelse(comp_2_TP$player_A == x, comp_2_TP$player_B, comp_2_TP$player_A)
comp_2_TP
################ON "SORT" LES JOUEURS DES QUATRE LINEUPS IMPLIQUEES DANS CHAQUE COMPARAISON
for(i in 1:nrow_comp_2_TP) {
##PA_T_l_players
comp_2_TP$PA_T_l_p1[i] <- strsplit(comp_2_TP$PA_T_l_players, ", ")[[i]][1]
comp_2_TP$PA_T_l_p2[i] <- strsplit(comp_2_TP$PA_T_l_players, ", ")[[i]][2]
comp_2_TP$PA_T_l_p3[i] <- strsplit(comp_2_TP$PA_T_l_players, ", ")[[i]][3]
comp_2_TP$PA_T_l_p4[i] <- strsplit(comp_2_TP$PA_T_l_players, ", ")[[i]][4]
comp_2_TP$PA_T_l_p5[i] <- strsplit(comp_2_TP$PA_T_l_players, ", ")[[i]][5]
##PB_T_l_players
comp_2_TP$PB_T_l_p1[i] <- strsplit(comp_2_TP$PB_T_l_players, ", ")[[i]][1]
comp_2_TP$PB_T_l_p2[i] <- strsplit(comp_2_TP$PB_T_l_players, ", ")[[i]][2]
comp_2_TP$PB_T_l_p3[i] <- strsplit(comp_2_TP$PB_T_l_players, ", ")[[i]][3]
comp_2_TP$PB_T_l_p4[i] <- strsplit(comp_2_TP$PB_T_l_players, ", ")[[i]][4]
comp_2_TP$PB_T_l_p5[i] <- strsplit(comp_2_TP$PB_T_l_players, ", ")[[i]][5]
##PA_O_l_players
comp_2_TP$PA_O_l_p1[i] <- strsplit(comp_2_TP$PA_O_l_players, ", ")[[i]][1]
comp_2_TP$PA_O_l_p2[i] <- strsplit(comp_2_TP$PA_O_l_players, ", ")[[i]][2]
comp_2_TP$PA_O_l_p3[i] <- strsplit(comp_2_TP$PA_O_l_players, ", ")[[i]][3]
comp_2_TP$PA_O_l_p4[i] <- strsplit(comp_2_TP$PA_O_l_players, ", ")[[i]][4]
comp_2_TP$PA_O_l_p5[i] <- strsplit(comp_2_TP$PA_O_l_players, ", ")[[i]][5]
##PB_O_l_players
comp_2_TP$PB_O_l_p1[i] <- strsplit(comp_2_TP$PB_O_l_players, ", ")[[i]][1]
comp_2_TP$PB_O_l_p2[i] <- strsplit(comp_2_TP$PB_O_l_players, ", ")[[i]][2]
comp_2_TP$PB_O_l_p3[i] <- strsplit(comp_2_TP$PB_O_l_players, ", ")[[i]][3]
comp_2_TP$PB_O_l_p4[i] <- strsplit(comp_2_TP$PB_O_l_players, ", ")[[i]][4]
comp_2_TP$PB_O_l_p5[i] <- strsplit(comp_2_TP$PB_O_l_players, ", ")[[i]][5]
}
comp_2_TP
comp_2_TP$PA_T_l_p1 <- as.numeric(comp_2_TP$PA_T_l_p1)
comp_2_TP$PA_T_l_p2 <- as.numeric(comp_2_TP$PA_T_l_p2)
comp_2_TP$PA_T_l_p3 <- as.numeric(comp_2_TP$PA_T_l_p3)
comp_2_TP$PA_T_l_p4 <- as.numeric(comp_2_TP$PA_T_l_p4)
comp_2_TP$PA_T_l_p5 <- as.numeric(comp_2_TP$PA_T_l_p5)
comp_2_TP$PB_T_l_p1 <- as.numeric(comp_2_TP$PB_T_l_p1)
comp_2_TP$PB_T_l_p2 <- as.numeric(comp_2_TP$PB_T_l_p2)
comp_2_TP$PB_T_l_p3 <- as.numeric(comp_2_TP$PB_T_l_p3)
comp_2_TP$PB_T_l_p4 <- as.numeric(comp_2_TP$PB_T_l_p4)
comp_2_TP$PB_T_l_p5 <- as.numeric(comp_2_TP$PB_T_l_p5)
comp_2_TP$PA_O_l_p1 <- as.numeric(comp_2_TP$PA_O_l_p1)
comp_2_TP$PA_O_l_p2 <- as.numeric(comp_2_TP$PA_O_l_p2)
comp_2_TP$PA_O_l_p3 <- as.numeric(comp_2_TP$PA_O_l_p3)
comp_2_TP$PA_O_l_p4 <- as.numeric(comp_2_TP$PA_O_l_p4)
comp_2_TP$PA_O_l_p5 <- as.numeric(comp_2_TP$PA_O_l_p5)
comp_2_TP$PB_O_l_p1 <- as.numeric(comp_2_TP$PB_O_l_p1)
comp_2_TP$PB_O_l_p2 <- as.numeric(comp_2_TP$PB_O_l_p2)
comp_2_TP$PB_O_l_p3 <- as.numeric(comp_2_TP$PB_O_l_p3)
comp_2_TP$PB_O_l_p4 <- as.numeric(comp_2_TP$PB_O_l_p4)
comp_2_TP$PB_O_l_p5 <- as.numeric(comp_2_TP$PB_O_l_p5)
################ON ASSOCIE A "TP_ID" et "CP_ID" LES JOUEURS DE LEUR LINEUP ET CEUX DE LA LINEUP EN FACE
comp_2_TP$TP_T_l_players <- ifelse(comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_T_l_players, comp_2_TP$PB_T_l_players)
comp_2_TP$CP_T_l_players <- ifelse(comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_T_l_players, comp_2_TP$PB_T_l_players)
comp_2_TP$TP_O_l_players <- ifelse(comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_O_l_players, comp_2_TP$PB_O_l_players)
comp_2_TP$CP_O_l_players <- ifelse(comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_O_l_players, comp_2_TP$PB_O_l_players)
comp_2_TP
comp_2_TP$TP_T_l <- ifelse(comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_T_l, comp_2_TP$PB_T_l)
comp_2_TP$CP_T_l <- ifelse(comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_T_l, comp_2_TP$PB_T_l)
comp_2_TP$TP_O_l <- ifelse(comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_O_l, comp_2_TP$PB_O_l)
comp_2_TP$CP_O_l <- ifelse(comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_O_l, comp_2_TP$PB_O_l)
comp_2_TP
comp_2_TP$TP_T_l_O_l_occ <- ifelse(comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$TP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_T_l_O_l_occ, comp_2_TP$PB_T_l_O_l_occ)
comp_2_TP$CP_T_l_O_l_occ <- ifelse(comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p1 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p2 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p3 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p4 | comp_2_TP$CP_ID == comp_2_TP$PA_T_l_p5, comp_2_TP$PA_T_l_O_l_occ, comp_2_TP$PB_T_l_O_l_occ)
comp_2_TP
################DANS LE TABLEAU DES COMPARAISONS ("comp_2_TP"), ON CREE LES COLONNES DANS QUI VONT PERMETTRE LES COMPARAISONS
TP_raw <- data_raw[[k]]
TP_raw
nrow_TP_raw <- nrow(TP_raw)
nrow_TP_raw
ncol_TP_raw <- ncol(TP_raw)
ncol_TP_raw
for(i in 1:nrow_comp_2_TP)
for(j in 1:nrow_TP_raw)
TP_raw[j,i+ncol_TP_raw] <- ifelse((TP_raw$T_l[j] == comp_2_TP$TP_T_l[i] & TP_raw$O_l[j] == comp_2_TP$TP_O_l[i] & TP_raw$occ[j] == comp_2_TP$TP_T_l_O_l_occ[i]), comp_2_TP$TP_ID[i], ifelse((TP_raw$T_l[j] == comp_2_TP$CP_T_l[i] & TP_raw$O_l[j] == comp_2_TP$CP_O_l[i] & TP_raw$occ[j] == comp_2_TP$CP_T_l_O_l_occ[i]), comp_2_TP$CP_ID[i], 0) )
TP_raw
####NOUVEL OBJET: H (MEME NOMBRE DE LIGNES QUE "comp_2_TP")
col1 <- match("V61", names(TP_raw))
col1
col_last <- ncol(TP_raw)
col_last
H1 <- data.frame(t(TP_raw[,c(col1:col_last)]))
H1
nrow_H1 <- nrow(H1)
ncol_H1 <- ncol(H1)
H2 <- H1
################
H2$TP_ID <- comp_2_TP$TP_ID
H2$CP_ID <- comp_2_TP$CP_ID
H2
for(i in 1:nrow_H1)
H2$TP_n[i] <- length(which(as.numeric(as.vector(H1[i,])) == H2$TP_ID[i]))
for(i in 1:nrow_H1)
H2$CP_n[i] <- length(which(as.numeric(as.vector(H1[i,])) == H2$CP_ID[i]))
H2
################ON ISOLE LES DIFFERENTES MESURES
##
H_StartTime <- as.character(TP_raw$StartTime)
H_StartTime
##
H_EndTime <- as.character(TP_raw$EndTime)
H_EndTime
##
H_ElapsedTime <- TP_raw$ElapsedSecs
H_ElapsedTime
##
H_Possessions_T <- if(TP_home_away == "Home") {as.vector(TP_raw$PossessionsHome)} else {as.vector(TP_raw$PossessionsAway)}
H_Possessions_T
H_Possessions_O <- if(TP_home_away == "Home") {as.vector(TP_raw$PossessionsAway)} else {as.vector(TP_raw$PossessionsHome)}
H_Possessions_O
##
H_PointsScored_T <- if(TP_home_away == "Home") {as.vector(TP_raw$PointsScoredHome)} else {as.vector(TP_raw$PointsScoredAway)}
H_PointsScored_T
H_PointsScored_O <- if(TP_home_away == "Home") {as.vector(TP_raw$PointsScoredAway)} else {as.vector(TP_raw$PointsScoredHome)}
H_PointsScored_O
##
H_PlusMinus <- if(TP_home_away == "Home") {as.vector(TP_raw$PlusMinusHome)} else {as.vector(TP_raw$PlusMinusAway)}
H_PlusMinus
################ON CALCULE LES DIFFERENTES MESURES POUR TP ET CP
######START TIME
##TP
H_StartTime_TP <- H1
H_StartTime_TP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_StartTime_TP[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_StartTime[j], 0)
H_StartTime_TP
for(i in 1:nrow_H1)
H_StartTime_TP$x2[i] <- H_StartTime[which(H1[i,] %in% H2$TP_ID[i])]
##CP
H_StartTime_CP <- H1
H_StartTime_CP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_StartTime_CP[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_StartTime[j], 0)
H_StartTime_CP
for(i in 1:nrow_H1)
H_StartTime_CP$x2[i] <- H_StartTime[which(H1[i,] %in% H2$CP_ID[i])]
##
H2$TP_StartTime <- H_StartTime_TP$x2
H2$CP_StartTime <- H_StartTime_CP$x2
H2
######END TIME
##TP
H_EndTime_TP <- H1
H_EndTime_TP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_EndTime_TP[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_EndTime[j], 0)
H_EndTime_TP
for(i in 1:nrow_H1)
H_EndTime_TP$x2[i] <- H_EndTime[which(H1[i,] %in% H2$TP_ID[i])]
##CP
H_EndTime_CP <- H1
H_EndTime_CP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_EndTime_CP[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_EndTime[j], 0)
H_EndTime_CP
for(i in 1:nrow_H1)
H_EndTime_CP$x2[i] <- H_EndTime[which(H1[i,] %in% H2$CP_ID[i])]
##
H2$TP_EndTime <- H_EndTime_TP$x2
H2$CP_EndTime <- H_EndTime_CP$x2
H2
######ELAPSED TIME
##TP
H_ElapsedTime_TP <- H1
H_ElapsedTime_TP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_ElapsedTime_TP[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_ElapsedTime[j], 0)
H_ElapsedTime_TP
H_ElapsedTime_TP$x1 <- rowSums(H_ElapsedTime_TP)
H_ElapsedTime_TP
#AUTRE METHODE
for(i in 1:nrow_H1)
H_ElapsedTime_TP$x2[i] <- H_ElapsedTime[which(H1[i,] %in% H2$TP_ID[i])]
##CP
H_ElapsedTime_CP <- H1
H_ElapsedTime_CP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_ElapsedTime_CP[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_ElapsedTime[j], 0)
H_ElapsedTime_CP
H_ElapsedTime_CP$x1 <- rowSums(H_ElapsedTime_CP)
H_ElapsedTime_CP
#AUTRE METHODE
for(i in 1:nrow_H1)
H_ElapsedTime_CP$x2[i] <- H_ElapsedTime[which(H1[i,] %in% H2$CP_ID[i])]
##
H2$TP_ElapsedTime <- H_ElapsedTime_TP$x2
H2$CP_ElapsedTime <- H_ElapsedTime_CP$x2
H2
######POSSESSIONS
##TP_T
H_Possessions_TP_T <- H1
H_Possessions_TP_T
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_Possessions_TP_T[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_Possessions_T[j], 0)
H_Possessions_TP_T
H_Possessions_TP_T$x1 <- rowSums(H_Possessions_TP_T)
H_Possessions_TP_T
#AUTRE METHODE
for(i in 1:nrow_H1)
H_Possessions_TP_T$x2[i] <- H_Possessions_T[which(H1[i,] %in% H2$TP_ID[i])]
##TP_O
H_Possessions_TP_O <- H1
H_Possessions_TP_O
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_Possessions_TP_O[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_Possessions_O[j], 0)
H_Possessions_TP_O
H_Possessions_TP_O$x1 <- rowSums(H_Possessions_TP_O)
H_Possessions_TP_O
#AUTRE METHODE
for(i in 1:nrow_H1)
H_Possessions_TP_O$x2[i] <- H_Possessions_O[which(H1[i,] %in% H2$TP_ID[i])]
##CP_T
H_Possessions_CP_T <- H1
H_Possessions_CP_T
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_Possessions_CP_T[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_Possessions_T[j], 0)
H_Possessions_CP_T
H_Possessions_CP_T$x1 <- rowSums(H_Possessions_CP_T)
H_Possessions_CP_T
#AUTRE METHODE
for(i in 1:nrow_H1)
H_Possessions_CP_T$x2[i] <- H_Possessions_T[which(H1[i,] %in% H2$CP_ID[i])]
##CP_O
H_Possessions_CP_O <- H1
H_Possessions_CP_O
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_Possessions_CP_O[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_Possessions_O[j], 0)
H_Possessions_CP_O
H_Possessions_CP_O$x1 <- rowSums(H_Possessions_CP_O)
H_Possessions_CP_O
#AUTRE METHODE
for(i in 1:nrow_H1)
H_Possessions_CP_O$x2[i] <- H_Possessions_O[which(H1[i,] %in% H2$CP_ID[i])]
##
H2$TP_Possessions_T <- H_Possessions_TP_T$x2
H2$TP_Possessions_O <- H_Possessions_TP_O$x2
H2$CP_Possessions_T <- H_Possessions_CP_T$x2
H2$CP_Possessions_O <- H_Possessions_CP_O$x2
H2
######POINTS SCORED
##TP_T
H_PointsScored_TP_T <- H1
H_PointsScored_TP_T
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_PointsScored_TP_T[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_PointsScored_T[j], 0)
H_PointsScored_TP_T
H_PointsScored_TP_T$x1 <- rowSums(H_PointsScored_TP_T)
H_PointsScored_TP_T
#AUTRE METHODE
for(i in 1:nrow_H1)
H_PointsScored_TP_T$x2[i] <- H_PointsScored_T[which(H1[i,] %in% H2$TP_ID[i])]
##TP_O
H_PointsScored_TP_O <- H1
H_PointsScored_TP_O
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_PointsScored_TP_O[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_PointsScored_O[j], 0)
H_PointsScored_TP_O
H_PointsScored_TP_O$x1 <- rowSums(H_PointsScored_TP_O)
H_PointsScored_TP_O
#AUTRE METHODE
for(i in 1:nrow_H1)
H_PointsScored_TP_O$x2[i] <- H_PointsScored_O[which(H1[i,] %in% H2$TP_ID[i])]
##CP_T
H_PointsScored_CP_T <- H1
H_PointsScored_CP_T
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_PointsScored_CP_T[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_PointsScored_T[j], 0)
H_PointsScored_CP_T
H_PointsScored_CP_T$x1 <- rowSums(H_PointsScored_CP_T)
H_PointsScored_CP_T
#AUTRE METHODE
for(i in 1:nrow_H1)
H_PointsScored_CP_T$x2[i] <- H_PointsScored_T[which(H1[i,] %in% H2$CP_ID[i])]
##CP_O
H_PointsScored_CP_O <- H1
H_PointsScored_CP_O
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_PointsScored_CP_O[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_PointsScored_O[j], 0)
H_PointsScored_CP_O
H_PointsScored_CP_O$x1 <- rowSums(H_PointsScored_CP_O)
H_PointsScored_CP_O
#AUTRE METHODE
for(i in 1:nrow_H1)
H_PointsScored_CP_O$x2[i] <- H_PointsScored_O[which(H1[i,] %in% H2$CP_ID[i])]
##
H2$TP_PointsScored_T <- H_PointsScored_TP_T$x2
H2$TP_PointsScored_O <- H_PointsScored_TP_O$x2
H2$CP_PointsScored_T <- H_PointsScored_CP_T$x2
H2$CP_PointsScored_O <- H_PointsScored_CP_O$x2
H2
######PLUS/MINUS RAW
##TP
H_PlusMinus_TP <- H1
H_PlusMinus_TP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_PlusMinus_TP[i,j] <- ifelse(H2[i,j] == H2$TP_ID[i], H_PlusMinus[j], 0)
H_PlusMinus_TP
H_PlusMinus_TP$x1 <- rowSums(H_PlusMinus_TP)
H_PlusMinus_TP
#AUTRE METHODE
for(i in 1:nrow_H1)
H_PlusMinus_TP$x2[i] <- H_PlusMinus[which(H1[i,] %in% H2$TP_ID[i])]
##CP
H_PlusMinus_CP <- H1
H_PlusMinus_CP
for(i in 1:nrow_H1)
for(j in 1:ncol_H1)
H_PlusMinus_CP[i,j] <- ifelse(H2[i,j] == H2$CP_ID[i], H_PlusMinus[j], 0)
H_PlusMinus_CP
H_PlusMinus_CP$x1 <- rowSums(H_PlusMinus_CP)
H_PlusMinus_CP
#AUTRE METHODE
for(i in 1:nrow_H1)
H_PlusMinus_CP$x2[i] <- H_PlusMinus[which(H1[i,] %in% H2$CP_ID[i])]
##
H2$TP_PM_raw <- H_PlusMinus_TP$x2
H2$CP_PM_raw <- H_PlusMinus_CP$x2
######PLUS/MINUS PER-POSSESSION
H2$TP_PM_perposs <- (H2$TP_PointsScored_T/H2$TP_Possessions_T) - (H2$TP_PointsScored_O/H2$TP_Possessions_O)
H2$CP_PM_perposs <- (H2$CP_PointsScored_T/H2$CP_Possessions_T) - (H2$CP_PointsScored_O/H2$CP_Possessions_O)
################ON MET EN FORME LES RESULTATS
TP_results <- data.frame(as.vector(rep(c(gameID), nrow_H1)), as.vector(rep(c(date), nrow_H1)), as.vector(rep(c(team_name), nrow_H1)), as.vector(rep(c(opp_name), nrow_H1)), as.vector(rep(c(TP_home_away), nrow_H1)), H2$TP_ID, H2$CP_ID, comp_2_TP$TP_T_l, comp_2_TP$CP_T_l, comp_2_TP$TP_O_l, comp_2_TP$CP_O_l, comp_2_TP$TP_T_l_players, comp_2_TP$CP_T_l_players, comp_2_TP$TP_O_l_players, comp_2_TP$CP_O_l_players, H2$TP_n, H2$CP_n, H2$TP_StartTime, H2$TP_EndTime, H2$TP_ElapsedTime, H2$CP_StartTime, H2$CP_EndTime, H2$CP_ElapsedTime, H2$TP_Possessions_T, H2$TP_Possessions_O, H2$CP_Possessions_T, H2$CP_Possessions_O, H2$TP_PointsScored_T, H2$TP_PointsScored_O, H2$CP_PointsScored_T, H2$CP_PointsScored_O, H2$TP_PM_raw, H2$CP_PM_raw, H2$TP_PM_perposs, H2$CP_PM_perposs)
names(TP_results) <- c("gameID", "date", "team_name", "opp_name", "TP_Home/Away", "TP", "CP", "TP_team", "CP_team", "TP_opp", "CP_opp", "TP_team_players", "CP_team_players", "TP_opp_players", "CP_opp_players", "TP_n", "CP_n", "TP_StartTime", "TP_EndTime", "TP_ElapsedTime", "CP_StartTime", "CP_EndTime", "CP_ElapsedTime", "TP_Possessions_T", "TP_Possessions_O", "CP_Possessions_T", "CP_Possessions_O", "TP_PointsScored_T", "TP_PointsScored_O", "CP_PointsScored_T", "CP_PointsScored_O", "TP_PM_raw", "CP_PM_raw", "TP_PM_perposs", "CP_PM_perposs")
TP_results
TP_results_raw <- TP_results[order(TP_results$CP),]
TP_results_raw
##ALL THINGS BEING EQUAL ("ATBE")
TP_results_ATBE <- subset(TP_results_raw, TP_results_raw$TP_opp == TP_results_raw$CP_opp)
TP_results_ATBE
##
result[[k]] <- list(gameID = gameID, TP_home_away = TP_home_away, team_name = team_name, opp_name = opp_name, date = date, T_lineups = T_lineups, O_lineups = O_lineups, T_l_unique = T_l_unique, O_l_unique = O_l_unique, comp_all_raw = comp_all_raw, comp_all = comp_all, comp_2 = comp_2, comp_2_TP = comp_2_TP, TP_raw = TP_raw, H2 = H2, TP_results_raw = TP_results_raw, TP_results_ATBE = TP_results_ATBE)
}
}
##BILAN GAMES SANS COMPARAISON "1"
w1 <- NULL
for (k in 1:length(game_ID))
if(is.atomic(result[[k]]) == TRUE) {w1[k] <- 1} else {w1[k] <- 0}
W <- data.frame(game_ID, w1)
W
sum(w1) ##nombre de games non pris en compte suivant ce critère
no_comparison_at_all <- which(W$w1 == 1)
no_comparison_at_all
##BILAN GAMES AVEC COMPARAISON "1" MAIS SANS COMPARAISON ATBE
w2 <- NULL
for (k in 1:length(game_ID))
if(k %in% no_comparison_at_all == FALSE) {w2[k] <- nrow(result[[k]]$TP_results_ATBE)} else {w2[k] <- NA}
W <- data.frame(game_ID, w1, w2)
W
w3 <- NULL
w4 <- NULL
for (k in 1:length(game_ID)) {
if(W$w2[k] == 0 & is.na(W$w2[k]) == FALSE) {w3[k] <- 1} else {w3[k] <- 0}
if(W$w2[k] != 0 & is.na(W$w2[k]) == FALSE) {w4[k] <- W$w2[k]} else {w4[k] <- 0}
}
W <- data.frame(game_ID, w1, w2, w3, w4)
W
sum(w3) ##nombre de games non pris en compte suivant ce critère
no_comparison_ATBE <- which(W$w3 == 1)
no_comparison_ATBE
sum(w4) ##nombre de comparaisons ATBE au final
##
results_ATBE_raw <- list()
for (k in 1:length(game_ID))
if(k %in% no_comparison_at_all == FALSE & k %in% no_comparison_ATBE == FALSE) {results_ATBE_raw[[k]] <- result[[k]]$TP_results_ATBE} else {results_ATBE_raw[[k]] <- NA}
results_ATBE_raw
results_ATBE_raw_OK <- do.call(rbind, results_ATBE_raw[!is.na(results_ATBE_raw)])
results_ATBE_raw_OK
nrow(results_ATBE_raw_OK)
##ON RETIRE LES COMPARAISONS IMPLIQUANT UN NOMBRE DE POSSESSIONS = 0 (PM = NaN ou Inf)
results_ATBE_OK <- subset(results_ATBE_raw_OK, results_ATBE_raw_OK$TP_Possessions_T != 0 & results_ATBE_raw_OK$TP_Possessions_O != 0 & results_ATBE_raw_OK$CP_Possessions_T != 0 & results_ATBE_raw_OK$CP_Possessions_O != 0)
####################
return(results_ATBE_OK)
}
############################## FUNCTION END #############################x <- qepm(8)
head(x)## gameID date team_name opp_name TP_Home/Away TP CP TP_team CP_team
## 3 14 20071101 PHX SEA Away 8 488 6 3
## 1 38 20071104 PHX CLE Home 8 16 5 1
## 7 38 20071104 PHX CLE Home 8 486 4 1
## 23 50 20071106 PHX CHA Away 8 488 8 3
## 50 75 20071109 PHX MIA Away 8 486 3 2
## 71 84 20071110 PHX ORL Away 8 488 7 4
## TP_opp CP_opp TP_team_players CP_team_players
## 3 2 2 16, 7, 4, 8, 537 488, 16, 7, 4, 537
## 1 8 8 7, 12, 486, 4, 8 16, 7, 12, 486, 4
## 7 8 8 16, 7, 12, 4, 8 16, 7, 12, 486, 4
## 23 12 12 16, 486, 4, 8, 78 488, 16, 486, 4, 78
## 50 4 4 16, 7, 12, 4, 8 16, 7, 12, 486, 4
## 71 8 8 16, 12, 4, 8, 78 488, 16, 12, 4, 78
## TP_opp_players CP_opp_players TP_n CP_n TP_StartTime
## 3 172, 714, 715, 168, 173 172, 714, 715, 168, 173 1 1 00:32:55
## 1 561, 257, 260, 264, 265 561, 257, 260, 264, 265 1 1 00:00:42
## 7 561, 257, 260, 264, 265 561, 257, 260, 264, 265 1 1 00:15:16
## 23 307, 126, 117, 90, 111 307, 126, 117, 90, 111 1 1 00:31:56
## 50 243, 201, 206, 185, 196 243, 201, 206, 185, 196 1 1 00:41:00
## 71 245, 277, 419, 149, 170 245, 277, 419, 149, 170 1 1 00:15:38
## TP_EndTime TP_ElapsedTime CP_StartTime CP_EndTime CP_ElapsedTime
## 3 00:31:50 65 00:33:18 00:32:55 23
## 1 00:00:00 42 00:14:46 00:12:00 166
## 7 00:14:46 30 00:14:46 00:12:00 166
## 23 00:30:40 76 00:32:44 00:31:56 48
## 50 00:39:06 114 00:15:18 00:14:58 20
## 71 00:14:42 56 00:14:42 00:12:00 162
## TP_Possessions_T TP_Possessions_O CP_Possessions_T CP_Possessions_O
## 3 3 2 1 1
## 1 2 2 7 6
## 7 2 2 7 6
## 23 3 3 2 3
## 50 4 4 1 1
## 71 2 2 5 5
## TP_PointsScored_T TP_PointsScored_O CP_PointsScored_T CP_PointsScored_O
## 3 0 2 0 3
## 1 2 3 7 5
## 7 3 3 7 5
## 23 2 0 0 4
## 50 3 4 0 0
## 71 4 0 2 4
## TP_PM_raw CP_PM_raw TP_PM_perposs CP_PM_perposs
## 3 -2 -3 -1.0000000 -3.0000000
## 1 -1 2 -0.5000000 0.1666667
## 7 0 2 0.0000000 0.1666667
## 23 2 -4 0.6666667 -1.3333333
## 50 -1 0 -0.2500000 0.0000000
## 71 4 -2 2.0000000 -0.4000000