add unit tests
This commit is contained in:
parent
3f8ab806e6
commit
b327147e83
512
.Rhistory
Normal file
512
.Rhistory
Normal file
@ -0,0 +1,512 @@
|
|||||||
|
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$kick_off)) {
|
||||||
|
data[is.na(data$kick_off$kick_off), ]$kick_off$kick_off <- FALSE
|
||||||
|
} else {
|
||||||
|
data$kick_off$kick_off <- FALSE
|
||||||
|
}
|
||||||
|
},
|
||||||
|
error = function(e) {
|
||||||
|
# handle the error
|
||||||
|
print(paste("An error occurred:", e$message))
|
||||||
|
})
|
||||||
|
data <- data %>% mutate(angle = atan(7.32 * x1 / (x1^2 + y1^2 - (7.32/2)^2)),
|
||||||
|
is_goal = ifelse(shot$outcome$id == 97, 1, 0)) %>%
|
||||||
|
select(-location)
|
||||||
|
data$shot$outcome <- data$shot$outcome %>% select(-id)
|
||||||
|
data <- data %>% unnest(shot, names_sep = "_")
|
||||||
|
data$angle <- ifelse(data$angle<0, base::pi + data$angle, data$angle)
|
||||||
|
data
|
||||||
|
}
|
||||||
|
data_test <- get_shots2("data/la_liga_events/303377.json")
|
||||||
|
View(data_test)
|
||||||
|
# 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$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
|
||||||
|
df_temp <- do.call(rbind, data$location)
|
||||||
|
colnames(df_temp) <- c("x1", "y1")
|
||||||
|
data$x1 <- df_temp[,1]
|
||||||
|
data$y1 <- df_temp[,2]
|
||||||
|
tryCatch({ # TODO reduce error cases
|
||||||
|
data$number_of_players <- mapply(function(sublist, x1_threshold) {
|
||||||
|
# Extracting the first location value and converting it to numeric
|
||||||
|
first_location_values <- sapply(sublist$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$kick_off)) {
|
||||||
|
data[is.na(data$kick_off$kick_off), ]$kick_off$kick_off <- FALSE
|
||||||
|
} else {
|
||||||
|
data$kick_off$kick_off <- FALSE
|
||||||
|
}
|
||||||
|
},
|
||||||
|
error = function(e) {
|
||||||
|
# handle the error
|
||||||
|
print(paste("An error occurred:", e$message))
|
||||||
|
})
|
||||||
|
data <- data %>% mutate(angle = atan(7.32 * x1 / (x1^2 + y1^2 - (7.32/2)^2)),
|
||||||
|
is_goal = ifelse(shot$outcome$id == 97, 1, 0)) %>%
|
||||||
|
select(-location)
|
||||||
|
data$angle <- ifelse(data$angle<0, base::pi + data$angle, data$angle)
|
||||||
|
data$shot$outcome <- data$shot$outcome %>% select(-id)
|
||||||
|
data <- data %>% unnest(shot, names_sep = "_")
|
||||||
|
%>% unnest(position, names_sep = "_")
|
||||||
|
# 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$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
|
||||||
|
df_temp <- do.call(rbind, data$location)
|
||||||
|
colnames(df_temp) <- c("x1", "y1")
|
||||||
|
data$x1 <- df_temp[,1]
|
||||||
|
data$y1 <- df_temp[,2]
|
||||||
|
tryCatch({ # TODO reduce error cases
|
||||||
|
data$number_of_players <- mapply(function(sublist, x1_threshold) {
|
||||||
|
# Extracting the first location value and converting it to numeric
|
||||||
|
first_location_values <- sapply(sublist$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$kick_off)) {
|
||||||
|
data[is.na(data$kick_off$kick_off), ]$kick_off$kick_off <- FALSE
|
||||||
|
} else {
|
||||||
|
data$kick_off$kick_off <- FALSE
|
||||||
|
}
|
||||||
|
},
|
||||||
|
error = function(e) {
|
||||||
|
# handle the error
|
||||||
|
print(paste("An error occurred:", e$message))
|
||||||
|
})
|
||||||
|
data <- data %>% mutate(angle = atan(7.32 * x1 / (x1^2 + y1^2 - (7.32/2)^2)),
|
||||||
|
is_goal = ifelse(shot$outcome$id == 97, 1, 0)) %>%
|
||||||
|
select(-location)
|
||||||
|
data$angle <- ifelse(data$angle<0, base::pi + data$angle, data$angle)
|
||||||
|
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 = "_")
|
||||||
|
data
|
||||||
|
}
|
||||||
|
data_test <- get_shots2("data/la_liga_events/303377.json")
|
||||||
|
View(data_test)
|
||||||
|
write.csv(data_test, file = "data/data_test.csv")
|
||||||
|
colnames(data_test)
|
||||||
|
data_test$kick_off
|
||||||
|
data_test$shot_body_part
|
||||||
|
data_df <- as.data.frame(data_test)
|
||||||
|
# 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$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
|
||||||
|
df_temp <- do.call(rbind, data$location)
|
||||||
|
colnames(df_temp) <- c("x1", "y1")
|
||||||
|
data$x1 <- df_temp[,1]
|
||||||
|
data$y1 <- df_temp[,2]
|
||||||
|
tryCatch({ # TODO reduce error cases
|
||||||
|
data$number_of_players <- mapply(function(sublist, x1_threshold) {
|
||||||
|
# Extracting the first location value and converting it to numeric
|
||||||
|
first_location_values <- sapply(sublist$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$message))
|
||||||
|
})
|
||||||
|
data <- data %>% mutate(angle = atan(7.32 * x1 / (x1^2 + y1^2 - (7.32/2)^2)),
|
||||||
|
is_goal = ifelse(shot$outcome$id == 97, 1, 0)) %>%
|
||||||
|
select(-location)
|
||||||
|
data$angle <- ifelse(data$angle<0, base::pi + data$angle, data$angle)
|
||||||
|
data$shot$outcome <- data$shot$outcome %>% select(-id)
|
||||||
|
data <- data %>% unnest(shot, names_sep = "_") %>%
|
||||||
|
unnest(position, names_sep = "_") %>%
|
||||||
|
unnest(shot_type, names_sep = "_") %>%
|
||||||
|
unnest(shot_outcome, names_sep = "_") %>%
|
||||||
|
unnest(shot_technique, names_sep = "_") %>%
|
||||||
|
unnest(shot_body_part, names_sep = "_")
|
||||||
|
data
|
||||||
|
}
|
||||||
|
data_test <- get_shots2("data/la_liga_events/303377.json")
|
||||||
|
write.csv(data_test, file = "data/data_test.csv")
|
||||||
|
file_names <- list.files(path = "data/la_liga_events/", pattern = "*.json")
|
||||||
|
data_list <- lapply(paste("data/la_liga_events/", file_names, sep = ""), get_shots2)
|
||||||
|
combined_data <- do.call(rbind, data_list)
|
||||||
|
data_test <- data <- fromJSON(("data/la_liga_events/303377.json") %>% filter(type$name == "Shot")
|
||||||
|
data_test <- fromJSON("data/la_liga_events/303377.json") %>% filter(type$name == "Shot")
|
||||||
|
data_test <- fromJSON("data/la_liga_events/303377.json") %>% filter(type$name == "Shot")
|
||||||
|
data_test$id
|
||||||
|
data_test$index
|
||||||
|
data4 <- read.csv("data/data4.csv", nrows = 1000)
|
||||||
|
View(data4)
|
||||||
|
# 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$name == "Shot") %>% dplyr::select(c(minute, position, location, shot))
|
||||||
|
df_temp <- do.call(rbind, data$location)
|
||||||
|
colnames(df_temp) <- c("x1", "y1")
|
||||||
|
data$x1 <- df_temp[,1]
|
||||||
|
data$y1 <- df_temp[,2]
|
||||||
|
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(angle = atan(7.32 * x1 / (x1^2 + y1^2 - (7.32/2)^2)),
|
||||||
|
is_goal = ifelse(shot$outcome$id == 97, 1, 0)) %>%
|
||||||
|
select(-location)
|
||||||
|
data$angle <- ifelse(data$angle<0, base::pi + data$angle, data$angle)
|
||||||
|
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
|
||||||
|
}
|
||||||
|
# sample data
|
||||||
|
data_test <- get_shots2("data/la_liga_events/303377.json")
|
||||||
|
write.csv(data_test, file = "data/data3_test.csv")
|
||||||
|
data1 <- read.csv("data/data1.csv", nrows = 1000)
|
||||||
|
colnames(data1)
|
||||||
|
View(data1)
|
||||||
|
data2 <- read.csv("data/data2.csv", nrows = 1000)
|
||||||
|
View(data2)
|
||||||
|
data3 <- read.csv("data/data3_test.csv", nrows = 1000)
|
||||||
|
colnames(data3)
|
||||||
|
View(data3)
|
||||||
BIN
data/.DS_Store
vendored
BIN
data/.DS_Store
vendored
Binary file not shown.
@ -53,16 +53,16 @@ get_shots <- function(file_path, name_detail, save_files = F){
|
|||||||
|
|
||||||
shots_ok
|
shots_ok
|
||||||
}
|
}
|
||||||
shotsEN <- get_shots("data/events/events_England.json", "EN")
|
# shotsEN <- get_shots("data/events/events_England.json", "EN")
|
||||||
shotsSP <- get_shots("data/events/events_Spain.json", "SP")
|
# shotsSP <- get_shots("data/events/events_Spain.json", "SP")
|
||||||
shotsWC <- get_shots("data/events/events_World_Cup.json", "WC")
|
# shotsWC <- get_shots("data/events/events_World_Cup.json", "WC")
|
||||||
shotsIT <- get_shots("data/events/events_Italy.json", "IT")
|
# shotsIT <- get_shots("data/events/events_Italy.json", "IT")
|
||||||
shotsGE <- get_shots("data/events/events_Germany.json", "GE")
|
# shotsGE <- get_shots("data/events/events_Germany.json", "GE")
|
||||||
shotsFR <- get_shots("data/events/events_France.json", "FR")
|
# shotsFR <- get_shots("data/events/events_France.json", "FR")
|
||||||
shotsEC <- get_shots("data/events/events_European_Championship.json", "EC")
|
# shotsEC <- get_shots("data/events/events_European_Championship.json", "EC")
|
||||||
|
#
|
||||||
shots <- shotsEN %>%
|
# shots <- shotsEN %>%
|
||||||
bind_rows(shotsFR, shotsGE, shotsIT, shotsSP, shotsWC, shotsEC)
|
# bind_rows(shotsFR, shotsGE, shotsIT, shotsSP, shotsWC, shotsEC)
|
||||||
|
|
||||||
get_final_data <- function(data) {
|
get_final_data <- function(data) {
|
||||||
|
|
||||||
@ -78,8 +78,8 @@ get_final_data <- function(data) {
|
|||||||
data
|
data
|
||||||
}
|
}
|
||||||
|
|
||||||
data1 <- get_final_data(shots)
|
# data1 <- get_final_data(shots)
|
||||||
write.csv(data1, file = "data/data1.csv")
|
# write.csv(data1, file = "data/data1.csv")
|
||||||
|
|
||||||
##################### The second dataset ##############################
|
##################### The second dataset ##############################
|
||||||
|
|
||||||
@ -96,8 +96,8 @@ get_data <- function(event_path, info_path) {
|
|||||||
|
|
||||||
}
|
}
|
||||||
|
|
||||||
data2 <- get_data(event_path = "data/events.csv", info_path = "data/ginf.csv")
|
# data2 <- get_data(event_path = "data/events.csv", info_path = "data/ginf.csv")
|
||||||
write.csv(data2, file = "data/data2.csv")
|
# write.csv(data2, file = "data/data2.csv")
|
||||||
|
|
||||||
# TODO dodać kolumne z x y coordinates bazując na location
|
# TODO dodać kolumne z x y coordinates bazując na location
|
||||||
# TODO dodać kolumnę z kątem do bramki
|
# TODO dodać kolumnę z kątem do bramki
|
||||||
@ -241,19 +241,17 @@ get_shots2 <- function(json_file) {
|
|||||||
data
|
data
|
||||||
}
|
}
|
||||||
|
|
||||||
file_names <- list.files(path = "data/la_liga_events/", pattern = "*.json")
|
# file_names <- list.files(path = "data/la_liga_events/", pattern = "*.json")
|
||||||
data_list <- lapply(paste("data/la_liga_events/", file_names, sep = ""), get_shots2)
|
# data_list <- lapply(paste("data/la_liga_events/", file_names, sep = ""), get_shots2)
|
||||||
combined_data <- do.call(rbind, data_list)
|
# combined_data <- do.call(rbind, data_list)
|
||||||
|
# ss <- lapply(data_list, nrow)
|
||||||
# sample data
|
# sss <- unlist(ss)
|
||||||
data_test <- get_shots2("data/la_liga_events/303377.json")
|
#
|
||||||
write.csv(data_test, file = "data/data3_test.csv")
|
# # sample data
|
||||||
|
# data_test <- get_shots2("data/la_liga_events/303377.json")
|
||||||
|
# write.csv(data_test, file = "data/data3_test.csv")
|
||||||
|
|
||||||
##################### The fourth dataset ##############################
|
##################### The fourth dataset ##############################
|
||||||
|
|
||||||
data4 <- read.csv("data/data4.csv", nrows = 1000)
|
|
||||||
data1 <- read.csv("data/data1.csv", nrows = 1000)
|
|
||||||
data2 <- read.csv("data/data2.csv", nrows = 1000)
|
|
||||||
data3 <- read.csv("data/data3_test.csv", nrows = 1000)
|
|
||||||
|
|
||||||
|
|
||||||
|
|||||||
32
testy/tests.r
Normal file
32
testy/tests.r
Normal file
@ -0,0 +1,32 @@
|
|||||||
|
library(tinytest)
|
||||||
|
|
||||||
|
# load all functions
|
||||||
|
source("notebooks/dataCleaning.R")
|
||||||
|
# set working directory on fantastyczne_gole
|
||||||
|
# setwd("~/Desktop/fantastyczne_gole")
|
||||||
|
# make sure that have all packages installed (from dataCleaning.R)
|
||||||
|
|
||||||
|
# generate random data
|
||||||
|
random_data <- as.data.frame(x = rnorm(100),
|
||||||
|
y = rnorm(100, mean = 1, sd = 3))
|
||||||
|
|
||||||
|
# unit tests
|
||||||
|
expect_silent(
|
||||||
|
data <- get_shots(file_path = "data/events/events_England.json", name_detail = "EN")
|
||||||
|
)
|
||||||
|
|
||||||
|
expect_silent(
|
||||||
|
data <- get_shots2(json_file = "data/la_liga_events/ (1).json")
|
||||||
|
)
|
||||||
|
|
||||||
|
expect_error(
|
||||||
|
get_final_data(data = random_data)
|
||||||
|
)
|
||||||
|
|
||||||
|
data1 <- get_shots2(json_file = "data/la_liga_events/ (1).json")
|
||||||
|
data2 <- get_shots2(json_file = "data/la_liga_events/ (2).json")
|
||||||
|
|
||||||
|
expect_length(colnames(data1), 22)
|
||||||
|
expect_length(colnames(data2), 22)
|
||||||
|
expect_false(is.vector(data1))
|
||||||
|
expect_true(is.data.frame(data1))
|
||||||
Loading…
Reference in New Issue
Block a user