Stratified split without repeated players of 18-19

PreparingData.R

@@ -192,14 +192,37 @@ for (name in playersIn2Clubs) {
 rownames(rowsPIn2C) <- playersIn2Clubs
 rowsPIn2C
 
+##############################################
+# Limpieza muestras erroneas temporada 18-19 #
+##############################################
+repeatedPercent <- c()
+variables <- paste(c("age.", "height.", "games.", "minutes.", "games_starts.",
+                     "xa.", "passes_pct.", "passes_total_distance.", "assisted_shots.", "passes_switches.", "pass_targets.", "passes_received.", "passes_received_pct.", "passes_pressure.",
+                     "xg.", "npxg.",  "shots_on_target_pct.", "goals_per_shot.", "pens_made.", "pens_att.", 
+                     "dribbles_completed_pct.", "goals_assists_per90.",
+                     "sca_passes_dead.", "gca_passes_dead.", "sca_passes_live.", "gca_passes_live.", "sca_dribbles.", "gca_dribbles.", "sca_fouled.", "gca_fouled.",
+                     "passes_intercepted.", "ball_recoveries.", "pressure_regains.", "fouls.", "tackles_won.", "tackles.", "aerials_won.", "aerials_lost.", "aerials_won_pct."), "1819", sep = "")
+for (i in c(1:4, 6:8)) {
+  player <- laLigaPlayers[rowsPIn2C[i,], ]
+  player <- player %>% select(all_of(c(variables, "player")))
+  percent <- mean(player[1, ] == player[2, ], na.rm = TRUE)
+  repeatedPercent <- rbind(repeatedPercent, cbind(player$player[1], percent))
+}
+repeatedPercent
+
+playersR.1819 <- laLigaPlayers[rowsPIn2C[c(1:4, 6:8),], ]
+playersR.1819 <- playersR.1819 %>% select(all_of(c("player", "squad.1819", variables)))
 
+# Ibai Gomez, Munir El Haddadi, Ruben Sobrino, Jeison Murillo, Facundo Roncaglia, Takashi Inui, Ruben Vezo
+# 4, 8, 13, 66, 72, 84, 109
+laLigaPlayers <- laLigaPlayers[-c(4, 8, 13, 66, 72, 109), ]
 ############################# ####################
 # CONJUNTO de ENTRENAMIENTO # # CONJUNTO de TEST #
 ############################# ####################
 set.seed(5682)
 library(rsample)
-jugadoresRepes <- laLigaPlayers[rowsPIn2C, ]
-laLigaPlayerSplit <- laLigaPlayers[-rowsPIn2C, ]
+jugadoresRepes <- laLigaPlayers[rowsPIn2C[-c(1:4, 6:8),], ]
+laLigaPlayerSplit <- laLigaPlayers[-rowsPIn2C[-c(1:4, 6:8),], ]
 car_split <- initial_split(laLigaPlayerSplit, prop = 8/10, strata = "position.1920")
 train <- training(car_split)
 train <- rbind(train, jugadoresRepes)
@@ -230,3 +253,6 @@ Youssef <- t(Youssef)
 # 7: Barcelona
 # 8: Sevilla
 
+jugador <- laLigaPlayers[rowsPIn2C[9,], ]
+jugador <- jugador %>% select(contains(variables))
+jugador <- t(jugador)