more data cleaning

notebooks/dataCleaning.R

@@ -117,6 +117,17 @@ loc2distance <- function(x, y) {
   sqrt(x^2 + y^2)
 }
 
+loc2locdistance <- function(x1, y1, x2, y2) {
+  sqrt( (x1 - x2)^2 + (y1 - y2)^2 )
+}
+
+# TODO
+# osobno bramkarz, wartość opponent_Goalkeeper
+# kolejność w row_number() wg odlegości do piłki z podziałem na teammate/opponent
+# przykład, najbliżej w teammate -> teammate_1,
+# przykład najbliżej w opponent -> opponent_1
+# w przypadku NA, wypełnić wartościami teammete_(nt+1):11, opponent_(no+1):11
+
 get_shots2 <- function(json_file) {
   data <- fromJSON(json_file) %>% filter(type$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
   
@@ -126,28 +137,68 @@ get_shots2 <- function(json_file) {
   data$x1 <- df_temp[,1]
   data$y1 <- df_temp[,2]
   
+  data$shot$freeze_frame <- Map(function(ff, x1, y1) {
+    ff$x1 <- yd_to_m(x1)
+    ff$y1 <- yd_to_m(y1)
+    return(ff)
+  }, 
+  data$shot$freeze_frame, data$x1, data$y1)
+
   tryCatch({
     df_players_location <- mapply( function(sublist) {
       if (!is.null(sublist$teammate)) {
         df_players <- sapply(sublist$location, function(loc) c(120, 80) - loc %>% as.numeric() %>% yd_to_m() %>% round(., digits = 1)) %>% t() %>% as.data.frame()
         # df <- sapply(sublist$teammate, function(tmt) cbind(df_players, tmt))
-        df <- cbind(df_players, sublist$teammate)
+        df <- cbind(df_players, sublist$teammate, sublist$position$name, sublist$x1, sublist$y1)
-        colnames(df) <- c("x", "y", "teammate")
+        colnames(df) <- c("x", "y", "teammate", "position_name", "x1", "y1")
+        df <- df %>% mutate(teammate = ifelse(teammate, "teammate", "opponent"),
+                            distance = loc2locdistance(x1 = x, y1 = y, x2 = x1, y2 = y1)) %>%
+                     arrange(distance)
+         
+        groups_count <- df %>% group_by(teammate) %>% count() %>% as.data.frame()
+        if ( !("opponent" %in% groups_count$teammate) ) {
+          groups_count <- groups_count %>% add_row(teammate = "opponent", n = 0)
+        } else if ( !("teammate" %in% groups_count$teammate) ) {
+          groups_count <- groups_count %>% add_row(teammate = "teammate", n = 0)
+        }
+        # df$distance <- loc2locdistance(x1 = df$x, y1 = df$y, x2 = data$x1, y1 = data$y1)
+        # df <- df %>% mutate(distance = loc2locdistance(x1 = x, y1 = y, x2 = data$x1, y1 = data$y1)) %>% arrange(distance)
+        # df <- df %>% arrange(distance)
         # df <- df %>% unnest(c(x, y, teammate))
         # df$x <- as.numeric(df$x)
         # df$y <- as.numeric(df$y)
         # df$teammate <- as.logical(df$teammate) %>% as.numeric()
-        # 
+        
-        na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
+        na_df <- as.data.frame(matrix("na", nrow = 21 - nrow(df), ncol = ncol(df)))
         colnames(na_df) <- colnames(df)
+        
+        na_df$teammate <- rep(c("opponent", "teammate"), c(11, 10) - groups_count$n)
         dff <- rbind(df, na_df)
+        dff <- dff %>% group_by(teammate) %>% mutate(rown = row_number(distance)) %>% ungroup() %>%
+          mutate(position_teammate = paste(teammate, ifelse(position_name == "Goalkeeper", position_name, rown), sep = "_")) %>%
+          select(-c(teammate, position_name, rown, distance, x1, y1)) %>%
+          mutate(x = ifelse(x == "na", NA, x),
+                 y = ifelse(x == "na", NA, y))
+        
+        # dff <- dff %>% mutate(row = row_number()) %>% 
+        #   mutate(position_teammate = paste(ifelse(position_name == "Goalkeeper", position_name, row), teammate, sep = "_")) %>% 
+        #   select(-c(teammate, position_name, row))
       } else {
-        dff <- as.data.frame(matrix(NA, nrow = 21, ncol = 3))
+        dff <- as.data.frame(matrix("na", nrow = 21, ncol = 3))
         colnames(dff) <- c("x", "y", "teammate")
+        dff$teammate <- rep(c("opponent", "teammate"), c(11, 10))
+        dff <- dff %>% group_by(teammate) %>% mutate(rown = row_number()) %>% ungroup() %>% 
+          mutate(position_teammate = paste(teammate, rown)) %>% 
+          select(c(-teammate, rown)) %>% 
+          mutate(x = ifelse(x == "na", NA, x),
+                 y = ifelse(x == "na", NA, y))
       }
+      # print(dff)
+      # print(wider_df)
+      # stop("123")
+      # %>%
       wider_df <- dff %>%
-        mutate(row = row_number()) %>%
+        pivot_wider(names_from = position_teammate, values_from = c(x, y), names_sep = "_player_")
-        pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
       # wider_df <- apply(wider_df, MARGIN = 2, unlist)
       wider_df
     }, data$shot$freeze_frame)
@@ -157,7 +208,9 @@ get_shots2 <- function(json_file) {
       print(json_file)
       print(paste("An error occurred:", e$message))
   })
+  print(df_players_location)
   df_players_location <- df_players_location %>% t()
+  print(df_players_location)
   
   # df <- do.call(rbind, df_test$shot$freeze_frame[[1]]$location) %>% as.data.frame()
   # dff <- cbind(df, df_test$shot$freeze_frame[[1]]$teammate)
@@ -313,7 +366,7 @@ combined_data <- bind_rows(data_list)
 # sss <- unlist(ss)
 # 
 # # sample data
-# data_test <- get_shots2("data/la_liga_events/ (1000).json")
+data <- fromJSON("data/la_liga_events/ (1000).json") %>% filter(type$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
 
 data3_final <- combined_data %>% select(-c(shot_outcome_name,
                                    shot_saved_off_target,