Data loading and univariate analysis separated in different files

PreparingData.R

+setwd("/home/mariogt/TFGs/Estadistica/")
+
+#####################################
+#             FUNCIONES             #
+#####################################
+aceptaTildes <- function(datos) {
+  datos$player <- gsub("É", "É", datos$player)
+  datos$player <- gsub("Á", "Á", datos$player) 
+  datos$player <- gsub("Ă“", "Ó", datos$player)
+  datos$player <- gsub("Ø", "Ø", datos$player)
+  datos$player <- gsub("ć", "ć", datos$player) 
+  datos$player <- gsub("š", "š", datos$player) 
+  datos$player <- gsub("Ă©", "é", datos$player) 
+  datos$player <- gsub("Ă«", "ë", datos$player) 
+  datos$player <- gsub("á", "á", datos$player) 
+  datos$player <- gsub("ĂŁ", "ã", datos$player) 
+  datos$player <- gsub("â", "â", datos$player) 
+  datos$player <- gsub("Ăł", "ó", datos$player)
+  datos$player <- gsub("Ăş", "ú", datos$player)
+  datos$player <- gsub("ñ", "ñ", datos$player)
+  datos$player <- gsub("Ă‘", "Ñ", datos$player)
+  datos$player <- gsub("ĂŻ", "ï", datos$player) 
+  datos$player <- gsub("Ă", "í", datos$player) 
+  datos$squad <- gsub("Ă©", "é", datos$squad)
+  return(datos)
+}
+
+######################################
+#          LECTURA de DATOS          #
+######################################
+allLeagues.1718 <- read.csv2("Datos/Kaggle/transfermarkt_fbref_201718.csv", header = TRUE, dec = ".")
+allLeagues.1819 <- read.csv2("Datos/Kaggle/transfermarkt_fbref_201819.csv", header = TRUE, dec = ".")
+allLeagues.1920 <- read.csv2("Datos/Kaggle/transfermarkt_fbref_201920.csv", header = TRUE, dec = ".")
+
+laLigaPlayersPos <- which(allLeagues.1718$league == "La Liga")
+laLigaPlayers.1718 <- allLeagues.1718[laLigaPlayersPos, ]
+dim(laLigaPlayers.1718)
+laLigaPlayers.1718 <- aceptaTildes(laLigaPlayers.1718)
+
+laLigaPlayersPos <- which(allLeagues.1819$league == "La Liga")
+laLigaPlayers.1819 <- allLeagues.1819[laLigaPlayersPos, ]
+dim(laLigaPlayers.1819)
+laLigaPlayers.1819 <- aceptaTildes(laLigaPlayers.1819)
+
+laLigaPlayersPos <- which(allLeagues.1920$league == "La Liga")
+laLigaPlayers.1920 <- allLeagues.1920[laLigaPlayersPos, ]
+dim(laLigaPlayers.1920)
+laLigaPlayers.1920 <- aceptaTildes(laLigaPlayers.1920)
+
+allLaLigaPlayer <- list(laLigaPlayers.1718, laLigaPlayers.1819, laLigaPlayers.1920)
+##########################################################
+# Jugadores que CAMBIARON de club en una misma temporada #
+##########################################################
+rowsPIn2CGlobal <- c()
+for (i in 1:length(allLaLigaPlayer)) {
+  repeatedPlayers <- which(duplicated(allLaLigaPlayer[[i]]$player) == TRUE)
+  playersIn2C <- allLaLigaPlayer[[i]]$player[repeatedPlayers]
+  season <- switch (i, "17-18", "18-19", "19-20")
+  for (name in playersIn2C) {
+    rows <- which(allLaLigaPlayer[[i]]$player == name)
+    teams <- c(allLaLigaPlayer[[i]]$squad[rows[1]], allLaLigaPlayer[[i]]$squad[rows[2]])
+    rowsPIn2CGlobal <- rbind(rowsPIn2CGlobal, c(name, rows, teams, season))
+  }
+}
+colnames(rowsPIn2CGlobal) <- c("Jugador", "Fila 1", "Fila 2", "Equipo 1º", "Equipo 2º", "Temporada")
+rowsPIn2CGlobal
+
+####################################################
+# Fusion de las muestras de jugadores en 2 equipos #
+####################################################.
+pos <- which(rowsPIn2CGlobal[,6] == "17-18")
+r1 <- as.numeric(rowsPIn2CGlobal[pos, 2])
+r2 <- as.numeric(rowsPIn2CGlobal[pos, 3])
+
+players <- c()
+for (i in 1:length(r1)) {
+  permanent <- laLigaPlayers.1718[r1[i], 1:12]
+  team1 <- as.numeric(laLigaPlayers.1718[r1[i], 13:400])
+  team2 <- as.numeric(laLigaPlayers.1718[r2[i], 13:400])
+  aux <- data.frame(rbind(team1, team2))
+  result <- apply(aux, 2, sum)
+  result <- c(permanent, result)
+  players <- rbind(players, result)
+}
+colnames(players) <- colnames(laLigaPlayers.1718)
+
+names <- rowsPIn2CGlobal[pos, 1]
+for (name in names) {
+  auxPos <- which(laLigaPlayers.1718$player == name)
+  laLigaPlayers.1718 <- laLigaPlayers.1718[-auxPos,]
+}
+laLigaPlayers.1718 <- rbind(laLigaPlayers.1718, players)
+
+pos <- which(rowsPIn2CGlobal[,6] == "18-19")
+r1 <- as.numeric(rowsPIn2CGlobal[pos, 2])
+r2 <- as.numeric(rowsPIn2CGlobal[pos, 3])
+
+players <- c()
+for (i in 1:length(r1)) {
+  permanent <- laLigaPlayers.1819[r1[i], 1:12]
+  team1 <- as.numeric(laLigaPlayers.1819[r1[i], 13:400])
+  team2 <- as.numeric(laLigaPlayers.1819[r2[i], 13:400])
+  aux <- data.frame(rbind(team1, team2))
+  result <- apply(aux, 2, sum)
+  result <- c(permanent, result)
+  players <- rbind(players, result)
+}
+colnames(players) <- colnames(laLigaPlayers.1819)
+
+names <- rowsPIn2CGlobal[pos, 1]
+for (name in names) {
+  auxPos <- which(laLigaPlayers.1819$player == name)
+  laLigaPlayers.1819 <- laLigaPlayers.1819[-auxPos,]
+}
+laLigaPlayers.1819 <- rbind(laLigaPlayers.1819, players)
+
+pos <- which(rowsPIn2CGlobal[,6] == "19-20")
+r1 <- as.numeric(rowsPIn2CGlobal[pos, 2])
+r2 <- as.numeric(rowsPIn2CGlobal[pos, 3])
+
+players <- c()
+for (i in 1:length(r1)) {
+  permanent <- laLigaPlayers.1920[r1[i], 1:12]
+  team1 <- as.numeric(laLigaPlayers.1920[r1[i], 13:400])
+  team2 <- as.numeric(laLigaPlayers.1920[r2[i], 13:400])
+  aux <- data.frame(rbind(team1, team2))
+  result <- apply(aux, 2, sum)
+  result <- c(permanent, result)
+  players <- rbind(players, result)
+}
+colnames(players) <- colnames(laLigaPlayers.1920)
+
+names <- rowsPIn2CGlobal[pos, 1]
+for (name in names) {
+  auxPos <- which(laLigaPlayers.1920$player == name)
+  laLigaPlayers.1920 <- laLigaPlayers.1920[-auxPos,]
+}
+laLigaPlayers.1920 <- rbind(laLigaPlayers.1920, players)
+
+##############
+# INNER JOIN #
+##############
+library(dplyr)
+laLigaPlayers.1719 <- inner_join(laLigaPlayers.1718, laLigaPlayers.1819, by = c("player" = "player"), suffix = c(".1718", ".1819"))
+auxLaLigaPlayers.1920 <- inner_join(laLigaPlayers.1920, laLigaPlayers.1920, by = c("player" = "player"), suffix = c(".1920", ".1920"))
+laLigaPlayers <- inner_join(laLigaPlayers.1719, auxLaLigaPlayers.1920, by = c("player" = "player"))
+
+# Veamos que jugadores jugaron en 2 CLUBES en algunas de las 3 temporadas contempladas
+repeatedPlayers <- which(duplicated(laLigaPlayers$player) == TRUE)
+playersIn2Clubs <- laLigaPlayers$player[repeatedPlayers]
+rowsPIn2C <- c()
+for (name in playersIn2Clubs) {
+  rowsPIn2C <- rbind(rowsPIn2C, which(laLigaPlayers$player == name)) 
+}
+rownames(rowsPIn2C) <- playersIn2Clubs
+rowsPIn2C
+
+
+############################# ####################
+# CONJUNTO de ENTRENAMIENTO # # CONJUNTO de TEST #
+############################# ####################
+set.seed(5682)
+ind <- sample(2, nrow(laLigaPlayers), replace = TRUE, prob = c(0.8, 0.2))
+train <- laLigaPlayers[ind == 1, ]
+test <- laLigaPlayers[ind == 2, ] 
\ No newline at end of file

UnivariateAnalysis.R

+setwd("/home/mariogt/TFGs/Estadistica/")
+
+#####################################
+#             FUNCIONES             #
+#####################################
+plotFor3Seasons <- function(data, variable) {
+  par(mfrow = c(1, 3))
+  dfs <- list()
+  maxPInPos <- 0
+  for (i in 1:3) {
+    season <- switch(i, ".1718", ".1819", ".1920")
+    var <- select(data, contains(paste(variable, season, sep = "")))
+    cat(names(var))
+    GK <- var %>% filter(str_detect(var[,1], 'GK'))
+    numGK <- dim(GK)[1]
+    DF <- var %>% filter(str_detect(var[,1], 'DF'))
+    numDF <- dim(DF)[1]
+    MF <- var %>% filter(str_detect(var[,1], 'MF'))
+    numMF <- dim(MF)[1]
+    FW <- var %>% filter(str_detect(var[,1], 'FW'))
+    numFW <- dim(FW)[1]
+    
+    df <- data.frame(
+      position = c("Portero", "Defensa", "Mediocentro", "Delantero"),
+      numberOf = c(numGK, numDF, numMF, numFW)
+    )
+    dfs[[i]] <- df
+    if(maxPInPos < max(df$numberOf)) {
+      maxPInPos <- max(df$numberOf)
+    }
+  }
+  for (i in 1:3) {
+    season <- switch(i, ".1718", ".1819", ".1920")
+    bp <- barplot(height = dfs[[i]]$numberOf, names = dfs[[i]]$position, 
+                  col = rgb(0.8,0.1,0.1,0.6),
+                  xlab = "posiciones", 
+                  ylab = "cantidad", 
+                  main= paste(c("Distribución posiciones ", season, "(TRAIN)"), sep = ""),
+                  ylim = c(0, maxPInPos)
+    ) 
+  }
+}
+
+univariantAnalysis <- function(data, variable, ids) {
+  info <- data %>% select(contains(paste(variable, ".", sep = "")))
+  info <- cbind(ids, info)
+  
+  # WIDE to LONG
+  dl <- gather(info, season, variable, 2:4, factor_key = TRUE)
+  dl[, 3] <- as.numeric(dl[, 3])
+  
+  bp <- ggplot(dl, aes(x = factor(season), y = variable)) +
+    geom_boxplot(alpha = 0.5, fill="darkblue", color="black", outlier.color="red")+
+    labs(title="Boxplot GCA por temporada", x = "Temporada", y = "Acciones de creación de gol")
+  
+  g <- group_by(dl, season)
+  g2 <- summarise(g, media = round(mean(variable), 1))
+  gp <- ggplot(g2, aes(x = season, media, group = 1))+
+    geom_point(alpha=0.5, color = "blue", size=3) +
+    geom_line(color = "red", size = 1) +
+    labs(title = "Promedio de GCA por temporada", x = "Temporada", y = "Acciones de creación de gol")
+  
+  figure <- ggarrange(bp, gp,
+                      labels = c("A", "B"),
+                      ncol = 1, nrow = 2)
+  figure
+}
+
+#############
+# LIBRERIAS #
+#############
+library(tidyverse)
+library(reshape)
+library(tidyr) # gather() -> WIDE data to LONG data
+library(ggplot2) # graphics
+library(ggpubr) # multiple ggplots 
+
+
+##########################################
+#          ANALISIS UNIVARIANTE          #
+##########################################
+
+# Get players identifiers
+ids <- data.frame(train$X)
+ids <- t(ids)
+      ########
+      # Edad #
+      ########
+      ##########
+      # Altura #
+      ##########
+      ################
+      # Pierna buena #
+      ################
+      ##################################
+      # Goles y asistencias por 90 min #
+      ##################################
+      ######
+      # xG #
+      ######
+      ########
+      # npxG #
+      ########
+      ######
+      # xA #
+      ######
+      ##################################
+      # Porcentaje de tiros a porteria #
+      ##################################
+      #######################
+      # Porcentaje de pases #
+      #######################
+
+passes_pct <- train %>% select(contains("passes_pct."))
+passes_pct <- cbind(ids, passes_pct)
+
+
+data_long = gather(passes_pct, season, passes, passes_pct.1718:passes_pct.1920, convert = TRUE, factor_key=TRUE)
+data_long
+data_long[, 3] <- as.numeric(data_long[, 3])
+
+
+ggplot(data_long, aes(x = factor(season), y = passes)) +
+  geom_boxplot(alpha = 0.5, fill="darkblue", color="black", outlier.color="red")+
+  labs(title="Boxplot susceptibilidad por tiempo y entrenamiento", x="Temporada",
+       y = "Porcentaje de pases exitosos")
+      ##################################
+      # Distancia total mediante pases #
+      ##################################
+      #############################
+      # Pases que derivan en tiro #
+      #############################
+      ###################################
+      # Cambios de orientación de juego #
+      ###################################
+      ###################
+      # SCA passes dead #
+      ###################
+sca_passes_dead <- train %>% select(contains("sca_passes_dead."))
+sca_passes_dead <- cbind(ids, sca_passes_dead)
+
+data_long = gather(sca_passes_dead, season, sca_passes_dead, sca_passes_dead.1718:sca_passes_dead.1920, convert = TRUE, factor_key=TRUE)
+data_long
+data_long[, 3] <- as.numeric(data_long[, 3])
+
+ggplot(data_long, aes(x = factor(season), y = sca_passes_dead)) +
+  geom_boxplot(alpha = 0.5, fill="darkblue", color="black", outlier.color="red")+
+  labs(title="Boxplot SCA por temporada", x="Temporada",
+       y = "Acciones de creación de tiro")
+
+      ###################
+      # GCA passes dead #
+      ###################
+
+
+univariantAnalysis(train, "passes_intercepted", ids)
+gca_passes_dead <- train %>% select(contains("gca_passes_dead."))
+gca_passes_dead <- cbind(ids, gca_passes_dead)
+par(mfrow = c(2, 1))
+data_long = gather(gca_passes_dead, season, gca_passes_dead, gca_passes_dead.1718:gca_passes_dead.1920, factor_key=TRUE)
+data_long
+data_long[, 3] <- as.numeric(data_long[, 3])
+
+ggplot(data_long, aes(x = factor(season), y = gca_passes_dead)) +
+  geom_boxplot(alpha = 0.5, fill="darkblue", color="black", outlier.color="red")+
+  labs(title="Boxplot GCA por temporada", x = "Temporada", y = "Acciones de creación de gol")
+
+g <- group_by(data_long, season)
+g2 <- summarise(g, media = round(mean(gca_passes_dead), 1))
+ggplot(g2, aes(x = season, media, group = 1))+
+  geom_point(alpha=0.5, color = "blue", size=3) +
+  geom_line(color = "red", size = 1) +
+  labs(title = "Promedio de GCA por temporada", x = "Temporada", y = "Acciones de creación de gol")
+      #######################
+      # Pases interceptados #
+      #######################
+      ####################################
+      # Porcentaje de presiones exitosas #
+      ####################################
+      ##################################
+      # Porcentaje de regates exitosos #
+      ##################################
+      ####################
+      # Faltas cometidas #
+      ####################
+      #######################################
+      # Porcentaje de juegos aereos ganados #
+      #######################################
+      ######################
+      # Penalties atajados #
+      ######################
+      ##############################
+      # Goles en contra por 90 min #
+      ##############################
+      ###################################
+      # Porcentaje de ocasiones paradas #
+      ###################################
+laLigaPlayers$foot <- as.factor(laLigaPlayers$foot)