res
}, data$shot$freeze_frame, data$x1)
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
tryCatch({ # TODO reduce error cases
data$number_of_players_teammates <- mapply(function(sublist, x1_threshold) {
# Extracting the first location value and converting it to numeric
first_location_values <- sapply(sublist$location, function(loc) as.numeric(loc[1]))
if ("teammate" %in% names(sublist)) {
# Filtering and counting
res <- sum(sublist$teammate & first_location_values > x1_threshold) # error here
} else {
res <- 0
}
res
}, data$shot$freeze_frame, data$x1)
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
data$shot <- data$shot %>% select(-freeze_frame, -statsbomb_xg, -key_pass_id)
data$shot$body_part <- data$shot$body_part %>% select(-id)
data$shot$technique <- data$shot$technique %>% select(-id)
data$shot$type <- data$shot$type %>% select(-id)
data$position <- data$position %>% select(-id)
data$shot <- data$shot %>% select(-end_location)
tryCatch({ # TODO reduce error cases
if ("one_on_one" %in% colnames(data$shot)) {
data[is.na(data$shot$one_on_one), ]$shot$one_on_one <- FALSE
} else {
data$shot$one_on_one <- FALSE
}
if ("first_time" %in% colnames(data$shot)) {
data[is.na(data$shot$first_time), ]$shot$first_time <- FALSE
} else {
data$shot$first_time <- FALSE
}
if ("aerial_won" %in% colnames(data$shot)) {
data[is.na(data$shot$aerial_won), ]$shot$aerial_won <- FALSE
} else {
data$shot$aerial_won <- FALSE
}
if ("saved_to_post" %in% colnames(data$shot)) {
data[is.na(data$shot$saved_to_post), ]$shot$saved_to_post <- FALSE
} else {
data$shot$saved_to_post <- FALSE
}
if ("deflected" %in% colnames(data$shot)) {
data[is.na(data$shot$deflected), ]$shot$deflected <- FALSE
} else {
data$shot$deflected <- FALSE
}
if ("saved_off_target" %in% colnames(data$shot)) {
data[is.na(data$shot$saved_off_target), ]$shot$saved_off_target <- FALSE
} else {
data$shot$saved_off_target <- FALSE
}
if ("open_goal" %in% colnames(data$shot)) {
data[is.na(data$shot$open_goal), ]$shot$open_goal <- FALSE
} else {
data$shot$open_goal <- FALSE
}
if ("follows_dribble" %in% colnames(data$shot)) {
data[is.na(data$shot$follows_dribble), ]$shot$follows_dribble <- FALSE
} else {
data$shot$follows_dribble <- FALSE
}
if ("redirect" %in% colnames(data$shot)) {
data[is.na(data$shot$redirect), ]$shot$redirect <- FALSE
} else {
data$shot$redirect <- FALSE
}
if ("kick_off" %in% colnames(data$kick_off)) {
data[is.na(data$shot$kick_off), ]$shotf$kick_off <- FALSE
} else {
data$kick_off <- FALSE
}
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
data <- data %>% mutate(is_goal = ifelse(shot$outcome$id == 97, 1, 0),
x1 = yd_to_m(x1) %>% round(., digits = 1),
y1 = yd_to_m(y1) %>% round(., digits = 1),
angle = loc2angle(x1, y1) %>% round(., digits = 1)) %>%
select(-location)
data$shot$outcome <- data$shot$outcome %>% select(-id)
data <- data %>% unnest(shot, names_sep = "_") %>%
unnest(position, names_sep = "_") %>%
unnest(shot_type, names_sep = "_") %>%
unnest(shot_outcome, names_sep = "_") %>%
unnest(shot_technique, names_sep = "_") %>%
unnest(shot_body_part, names_sep = "_")
data <- cbind(data, df_players_location)
data
}
data_list <- lapply(paste("data/la_liga_events/", file_names, sep = ""), get_shots2)
# TODO not all json data have these same keys/fields, customize function to cover all cases
get_shots2 <- function(json_file) {
data <- fromJSON(json_file) %>% filter(type$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
df_temp <- do.call(rbind, lapply(data$location, function(loc) c(120, 80) - loc))
colnames(df_temp) <- c("x1", "y1")
data$x1 <- df_temp[,1]
data$y1 <- df_temp[,2]
tryCatch({
df_players_location <- mapply( function(sublist) {
df_players <- sapply(sublist$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)
colnames(df) <- c("x", "y", "teammate")
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, data$shot$freeze_frame)
},
error = function(e) {
# handle the error
print(json_file)
print(paste("An error occurred:", e$message))
})
df_players_location <- df_players_location %>% t()
# df <- do.call(rbind, df_test$shot$freeze_frame[[1]]$location) %>% as.data.frame()
# dff <- cbind(df, df_test$shot$freeze_frame[[1]]$teammate)
# colnames(dff) <- c("x", "y", "teammate")
#
#
# na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(dff)))
# colnames(na_df) <- colnames(dff)
# dff <- rbind(dff, na_df)
#
# wider_df <- dff %>%
#   mutate(row = row_number()) %>%
#   pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
tryCatch({ # TODO reduce error cases
data$number_of_players_opponents <- mapply(function(sublist, x1_threshold) {
# Extracting the first location value and converting it to numeric
first_location_values <- sapply(sublist$location, function(loc) as.numeric(loc[1]))
if ("teammate" %in% names(sublist)) {
# Filtering and counting
res <- sum(!sublist$teammate & first_location_values > x1_threshold) # error here
} else {
res <- 0
}
res
}, data$shot$freeze_frame, data$x1)
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
tryCatch({ # TODO reduce error cases
data$number_of_players_teammates <- mapply(function(sublist, x1_threshold) {
# Extracting the first location value and converting it to numeric
first_location_values <- sapply(sublist$location, function(loc) as.numeric(loc[1]))
if ("teammate" %in% names(sublist)) {
# Filtering and counting
res <- sum(sublist$teammate & first_location_values > x1_threshold) # error here
} else {
res <- 0
}
res
}, data$shot$freeze_frame, data$x1)
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
data$shot <- data$shot %>% select(-freeze_frame, -statsbomb_xg, -key_pass_id)
data$shot$body_part <- data$shot$body_part %>% select(-id)
data$shot$technique <- data$shot$technique %>% select(-id)
data$shot$type <- data$shot$type %>% select(-id)
data$position <- data$position %>% select(-id)
data$shot <- data$shot %>% select(-end_location)
tryCatch({ # TODO reduce error cases
if ("one_on_one" %in% colnames(data$shot)) {
data[is.na(data$shot$one_on_one), ]$shot$one_on_one <- FALSE
} else {
data$shot$one_on_one <- FALSE
}
if ("first_time" %in% colnames(data$shot)) {
data[is.na(data$shot$first_time), ]$shot$first_time <- FALSE
} else {
data$shot$first_time <- FALSE
}
if ("aerial_won" %in% colnames(data$shot)) {
data[is.na(data$shot$aerial_won), ]$shot$aerial_won <- FALSE
} else {
data$shot$aerial_won <- FALSE
}
if ("saved_to_post" %in% colnames(data$shot)) {
data[is.na(data$shot$saved_to_post), ]$shot$saved_to_post <- FALSE
} else {
data$shot$saved_to_post <- FALSE
}
if ("deflected" %in% colnames(data$shot)) {
data[is.na(data$shot$deflected), ]$shot$deflected <- FALSE
} else {
data$shot$deflected <- FALSE
}
if ("saved_off_target" %in% colnames(data$shot)) {
data[is.na(data$shot$saved_off_target), ]$shot$saved_off_target <- FALSE
} else {
data$shot$saved_off_target <- FALSE
}
if ("open_goal" %in% colnames(data$shot)) {
data[is.na(data$shot$open_goal), ]$shot$open_goal <- FALSE
} else {
data$shot$open_goal <- FALSE
}
if ("follows_dribble" %in% colnames(data$shot)) {
data[is.na(data$shot$follows_dribble), ]$shot$follows_dribble <- FALSE
} else {
data$shot$follows_dribble <- FALSE
}
if ("redirect" %in% colnames(data$shot)) {
data[is.na(data$shot$redirect), ]$shot$redirect <- FALSE
} else {
data$shot$redirect <- FALSE
}
if ("kick_off" %in% colnames(data$kick_off)) {
data[is.na(data$shot$kick_off), ]$shotf$kick_off <- FALSE
} else {
data$kick_off <- FALSE
}
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
data <- data %>% mutate(is_goal = ifelse(shot$outcome$id == 97, 1, 0),
x1 = yd_to_m(x1) %>% round(., digits = 1),
y1 = yd_to_m(y1) %>% round(., digits = 1),
angle = loc2angle(x1, y1) %>% round(., digits = 1)) %>%
select(-location)
data$shot$outcome <- data$shot$outcome %>% select(-id)
data <- data %>% unnest(shot, names_sep = "_") %>%
unnest(position, names_sep = "_") %>%
unnest(shot_type, names_sep = "_") %>%
unnest(shot_outcome, names_sep = "_") %>%
unnest(shot_technique, names_sep = "_") %>%
unnest(shot_body_part, names_sep = "_")
data <- cbind(data, df_players_location)
data
}
data_list <- lapply(paste("data/la_liga_events/", file_names, sep = ""), get_shots2)
df <- fromJSON("data/la_liga_events/ (1006).json") %>% filter(type$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
df$shot$freeze_frame[[1]]$location
df$shot$freeze_frame[[1]]$teammate
mapply( function(sublist) {
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$teammate)
colnames(df) <- c("x", "y", "teammate")
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, df$shot$freeze_frame)
mapply( function(sublist) {
df_players <- sapply(sublist$location, function(loc) c(120, 80) - loc %>% as.numeric() %>% yd_to_m() %>% round(., digits = 1)) %>% t() %>% as.data.frame()
# df <- sapply(sublist$teammate, function(tmt) cbind(df_players, tmt))
print(df_players)
df <- cbind(df_players, sublist$teammate)
colnames(df) <- c("x", "y", "teammate")
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, df$shot$freeze_frame)
mapply( function(sublist) {
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))
print(df_players)
print(sublist$teammate)
df <- cbind(df_players, sublist$teammate)
colnames(df) <- c("x", "y", "teammate")
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, df$shot$freeze_frame)
df$shot$freeze_frame[[1]]$teammate
mapply( function(sublist) {
df_players <- sapply(sublist$location, function(loc) c(120, 80) - loc %>% as.numeric() %>% yd_to_m() %>% round(., digits = 1)) %>% t() %>% as.data.frame()
# df <- sapply(sublist$teammate, function(tmt) cbind(df_players, tmt))
print(df_players)
print(sublist$teammate)
df <- cbind(df_players, sublist$teammate)
colnames(df) <- c("x", "y", "teammate")
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, df$shot$freeze_frame)
mapply( function(sublist) {
df_players <- sapply(sublist$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)
colnames(df) <- c("x", "y", "teammate")
print(df)
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, df$shot$freeze_frame)
df$shot$freeze_frame[[3]]$teammate
df$shot$freeze_frame[[10]]$teammate
df$shot$freeze_frame[[115]]$teammate
df$shot$freeze_frame[[11]]$teammate
length(df$shot$freeze_frame)
df$shot$freeze_frame[[25]]$teammate
df$shot$freeze_frame[[24]]$teammate
# TODO not all json data have these same keys/fields, customize function to cover all cases
get_shots2 <- function(json_file) {
data <- fromJSON(json_file) %>% filter(type$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
df_temp <- do.call(rbind, lapply(data$location, function(loc) c(120, 80) - loc))
colnames(df_temp) <- c("x1", "y1")
data$x1 <- df_temp[,1]
data$y1 <- df_temp[,2]
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)
colnames(df) <- c("x", "y", "teammate")
na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(df)))
colnames(na_df) <- colnames(df)
dff <- rbind(df, na_df)
} else {
dff <- as.data.frame(matrix(NA, nrow = 21, ncol = 3))
colnames(dff) <- c("x", "y", "teammate")
}
wider_df <- dff %>%
mutate(row = row_number()) %>%
pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
wider_df
}, data$shot$freeze_frame)
},
error = function(e) {
# handle the error
print(json_file)
print(paste("An error occurred:", e$message))
})
df_players_location <- df_players_location %>% t()
# df <- do.call(rbind, df_test$shot$freeze_frame[[1]]$location) %>% as.data.frame()
# dff <- cbind(df, df_test$shot$freeze_frame[[1]]$teammate)
# colnames(dff) <- c("x", "y", "teammate")
#
#
# na_df <- as.data.frame(matrix(NA, nrow = 21 - nrow(df), ncol = ncol(dff)))
# colnames(na_df) <- colnames(dff)
# dff <- rbind(dff, na_df)
#
# wider_df <- dff %>%
#   mutate(row = row_number()) %>%
#   pivot_wider(names_from = row, values_from = c(x, y, teammate), names_sep = "_player")
tryCatch({ # TODO reduce error cases
data$number_of_players_opponents <- mapply(function(sublist, x1_threshold) {
# Extracting the first location value and converting it to numeric
first_location_values <- sapply(sublist$location, function(loc) as.numeric(loc[1]))
if ("teammate" %in% names(sublist)) {
# Filtering and counting
res <- sum(!sublist$teammate & first_location_values > x1_threshold) # error here
} else {
res <- 0
}
res
}, data$shot$freeze_frame, data$x1)
},
error = function(e) {
print(json_file)
# handle the error
print(paste("An error occurred:", e$message))
})
tryCatch({ # TODO reduce error cases
data$number_of_players_teammates <- mapply(function(sublist, x1_threshold) {
# Extracting the first location value and converting it to numeric
first_location_values <- sapply(sublist$location, function(loc) as.numeric(loc[1]))
if ("teammate" %in% names(sublist)) {
# Filtering and counting
res <- sum(sublist$teammate & first_location_values > x1_threshold) # error here
} else {
res <- 0
}
res
}, data$shot$freeze_frame, data$x1)
},
error = function(e) {
print(json_file)
# handle the error
print(paste("An error occurred:", e$message))
})
data$shot <- data$shot %>% select(-freeze_frame, -statsbomb_xg, -key_pass_id)
data$shot$body_part <- data$shot$body_part %>% select(-id)
data$shot$technique <- data$shot$technique %>% select(-id)
data$shot$type <- data$shot$type %>% select(-id)
data$position <- data$position %>% select(-id)
data$shot <- data$shot %>% select(-end_location)
tryCatch({ # TODO reduce error cases
if ("one_on_one" %in% colnames(data$shot)) {
data[is.na(data$shot$one_on_one), ]$shot$one_on_one <- FALSE
} else {
data$shot$one_on_one <- FALSE
}
if ("first_time" %in% colnames(data$shot)) {
data[is.na(data$shot$first_time), ]$shot$first_time <- FALSE
} else {
data$shot$first_time <- FALSE
}
if ("aerial_won" %in% colnames(data$shot)) {
data[is.na(data$shot$aerial_won), ]$shot$aerial_won <- FALSE
} else {
data$shot$aerial_won <- FALSE
}
if ("saved_to_post" %in% colnames(data$shot)) {
data[is.na(data$shot$saved_to_post), ]$shot$saved_to_post <- FALSE
} else {
data$shot$saved_to_post <- FALSE
}
if ("deflected" %in% colnames(data$shot)) {
data[is.na(data$shot$deflected), ]$shot$deflected <- FALSE
} else {
data$shot$deflected <- FALSE
}
if ("saved_off_target" %in% colnames(data$shot)) {
data[is.na(data$shot$saved_off_target), ]$shot$saved_off_target <- FALSE
} else {
data$shot$saved_off_target <- FALSE
}
if ("open_goal" %in% colnames(data$shot)) {
data[is.na(data$shot$open_goal), ]$shot$open_goal <- FALSE
} else {
data$shot$open_goal <- FALSE
}
if ("follows_dribble" %in% colnames(data$shot)) {
data[is.na(data$shot$follows_dribble), ]$shot$follows_dribble <- FALSE
} else {
data$shot$follows_dribble <- FALSE
}
if ("redirect" %in% colnames(data$shot)) {
data[is.na(data$shot$redirect), ]$shot$redirect <- FALSE
} else {
data$shot$redirect <- FALSE
}
if ("kick_off" %in% colnames(data$kick_off)) {
data[is.na(data$shot$kick_off), ]$shotf$kick_off <- FALSE
} else {
data$kick_off <- FALSE
}
},
error = function(e) {
# handle the error
print(paste("An error occurred:", e$message))
})
data <- data %>% mutate(is_goal = ifelse(shot$outcome$id == 97, 1, 0),
x1 = yd_to_m(x1) %>% round(., digits = 1),
y1 = yd_to_m(y1) %>% round(., digits = 1),
angle = loc2angle(x1, y1) %>% round(., digits = 1)) %>%
select(-location)
data$shot$outcome <- data$shot$outcome %>% select(-id)
data <- data %>% unnest(shot, names_sep = "_") %>%
unnest(position, names_sep = "_") %>%
unnest(shot_type, names_sep = "_") %>%
unnest(shot_outcome, names_sep = "_") %>%
unnest(shot_technique, names_sep = "_") %>%
unnest(shot_body_part, names_sep = "_")
data <- cbind(data, df_players_location)
data
}
data_list <- lapply(paste("data/la_liga_events/", file_names, sep = ""), get_shots2)
View(data3_final)
combined_data <- do.call(rbind, data_list)
data_list
combined_data <- do.call(rbind, data_list)
combined_data <- do.call(bind_rows, data_list)
nrow(combined_data)
data3_final <- combined_data %>% select(-c(shot_outcome_name,
shot_saved_off_target,
shot_saved_to_post)) %>%
mutate(is_head = ifelse(shot_body_part_name == "Head", 1, 0),
is_right_foot = ifelse(shot_body_part_name == "Right Foot", 1, 0),
is_left_foot = ifelse(shot_body_part_name == "Left Foot", 1, 0),
is_open_play = ifelse(shot_type_name == "Open Play", 1, 0)) %>%
select(-shot_body_part_name)
colnames(data3_final)
write.csv(data3_final, file = "data/final_data.csv")
write_csv(data3_final, file = "data/final_data.csv")