- STRUCT 과 UNNEST 처음 접해보는 내용이라, 복습 필요.
- PIVOT 내용 중 ANY_VALUE는 데이터 양이 많고, 어떤 데이터들이 어떤 특성을 가지고 담겨있는지 정확하게 모른다면 활용하면 위험하겠다는 생각이 들었음.
Q1. STRUCT, UNNEST
1. array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요.
SELECT title,
-- genres,
genre
FROM advanced.array_exercises AS ae
CROSS JOIN UNNEST(genres) AS genre
-- genres는 평면화가 된 데이터를 의미
-- genres가 지금 배열
-- ARRAY : 같은 타입의 여러 데이터를 저장하고 싶을 때
-- ARRAY를 풀때 Flattten(평면화) -> UNNEST
-- UNNEST릃 할 때는 CROSS JOIN + UNNEST(ARRAY_COLUMN) 컬럼 명시
2) array_exercises 테이블에서 각 영화(title)q별로 배우(actor)와 배역(character)을 보여주세요. 배우와 배역은 별도의 컬럼으로 나와야 함.
SELECT title,
-- actors
-- actor에 직접 접근하면 어떨까 -> 새로운 컬럼으로 가능하나, 매번 SAFE_OFFSET을 지정해야 함
-- actors = [STRUCT(STRING,STRING)]
actors[SAFE_OFFSET(0)].actor AS first_actor,
actors[SAFE_OFFSET(0)].actor AS first_character,
actors[SAFE_OFFSET(1)].actor AS second_actor,
actors[SAFE_OFFSET(1)].actor AS second_character
-- 배열에 직접 접근이 아닌 UNNEST로 풀어야 편리할 듯
FROM advanced.array_exercises as ae
---------------------------------------------------------------
---------------------------------------------------------------
SELECT title,
actor.actor,
actor.character
FROM advanced.array_exercises as ae
CROSS JOIN UNNEST(actors) AS actor
-- actors가 배열
3) array_exercises 테이블에서 각 영화(title) 별로 배우(actor), 배역(character), 장르(genre)를 출력. 한 row에 배우, 배역, 장르가 모두 표시되어야 함.
SELECT title,
-- actors, #ARRAY<STRUCT(STRING, STRING)>
actor.actor as actor,
actor.character as character,
-- genres # ARRAY<STRING>
genre
FROM advanced.array_exercises
CROSS JOIN UNNEST(actors) AS actor
CROSS JOIN UNNEST(genres) AS genre
Q2. PIVOT
1-1)
1) orders 테이블에서 유저(user_id)별로 주문 금액(amount)의 합계를 PIVOT해주세요.
날짜(order_date)를 행(Row)으로, user_id를 열(Column)으로 만들어야 합니다
SELECT order_date,
IF(user_id = 1, amount , NULL) AS user_1,
IF(user_id = 2, amount , NULL) AS user_2,
IF(user_id = 3, amount , NULL) AS user_3
FROM(
SELECT order_date,
user_id,
amount
FROM advanced.orders
GROUP BY order_date, user_id, amount
ORDER BY order_date
)
---------------------------------------------------------------
---------------------------------------------------------------
1-2)
SELECT order_date,
MAX(IF(user_id = 1, amount , NULL)) AS user_1,
MAX(IF(user_id = 2, amount , NULL)) AS user_2,
MAX(IF(user_id = 3, amount , NULL)) AS user_3
FROM(
SELECT order_date,
user_id,
amount
FROM advanced.orders
GROUP BY order_date, user_id, amount
ORDER BY order_date
)
GROUP BY order_date
ORDER BY order_date
---------------------------------------------------------------
---------------------------------------------------------------
2) orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT해주세요. user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다.
SELECT order_date,
SUM(IF(user_id = 1, amount , NULL)) AS user_1,
SUM(IF(user_id = 2, amount , NULL)) AS user_2,
SUM(IF(user_id = 3, amount , NULL)) AS user_3
FROM advanced.orders
GROUP BY order_date
ORDER BY order_date
backtick 활용
any value는 어디에 활용할 수 있을지? -> 데이터는 믿을수 없기에 일부 데이터만 보고 사용 판단하기엔 위험할 것 같음.
SELECT user_id,
SUM(IF(order_date = "2023-05-01", amount, 0)) AS `2023-05-01`,
SUM(IF(order_date = "2023-05-02", amount, 0)) AS `2023-05-02`,
SUM(IF(order_date = "2023-05-03", amount, 0)) AS `2023-05-03`,
SUM(IF(order_date = "2023-05-04", amount, 0)) AS `2023-05-04`,
SUM(IF(order_date = "2023-05-05", amount, 0)) AS `2023-05-05`,
FROM advanced.orders
GROUP BY user_id
ORDER BY user_id
---------------------------------------------------------------
---------------------------------------------------------------
3) orders 테이블에서 사용자(user_id)별, 날짜(order_date)별로 주문이 있다면 1, 없다면 0으로 PIVOT 해주세요. user_id를 행(Row)으로, order_date를 열(Column)로 만들고 주문을 많이 해도 1로 처리합니다
3-1) 주문 여부 1,0
SELECT user_id,
SUM(IF(order_date = "2023-05-01", 1, 0)) AS `2023-05-01`,
SUM(IF(order_date = "2023-05-02", 1, 0)) AS `2023-05-02`,
SUM(IF(order_date = "2023-05-03", 1, 0)) AS `2023-05-03`,
SUM(IF(order_date = "2023-05-04", 1, 0)) AS `2023-05-04`,
SUM(IF(order_date = "2023-05-05", 1, 0)) AS `2023-05-05`,
FROM advanced.orders
GROUP BY user_id
ORDER BY user_id
3-2) 횟수
SELECT user_id,
SUM(IF(order_date = "2023-05-01", 1, 0)) AS `2023-05-01`,
SUM(IF(order_date = "2023-05-02", 1, 0)) AS `2023-05-02`,
SUM(IF(order_date = "2023-05-03", 1, 0)) AS `2023-05-03`,
SUM(IF(order_date = "2023-05-04", 1, 0)) AS `2023-05-04`,
SUM(IF(order_date = "2023-05-05", 1, 0)) AS `2023-05-05`,
FROM advanced.orders
GROUP BY user_id
ORDER BY user_id
-- 앱 로그 PIVOT
WITH base AS(
SELECT event_date,
event_timestamp,
event_name,
user_id,
user_pseudo_id,
MAX(IF(param.key = "firebase_screen", param.value.string_value, NULL)) AS firebase_screen,
MAX(IF(param.key = "food_id", param.value.int_value, NULL)) AS food_id,
MAX(IF(param.key = "session_id", param.value.string_value, NULL)) AS session_id
-- * EXCEPT(event_params)
FROM advanced.app_logs
CROSS JOIN UNNEST(event_params) AS param
WHERE event_date = "2022-08-01"
GROUP BY ALL
)
SELECT event_date,
COUNT(user_id) AS user_cnt
FROM base
WHERE event_name = "click_cart" AND food_id = 1544
GROUP BY event_date
Q3. 퍼널
-- 이중 WITH 문
WITH BASE AS(
SELECT event_date,
event_timestamp,
event_name,
user_id,
user_pseudo_id,
platform,
-- event_param
MAX(IF(event_param.key = "firebase_screen", event_param.value.string_value, NULL )) AS firebase_screen,
-- MAX(IF(event_param.key = "food_id", event_param.value.int_value, NULL )) AS food_id,
MAX(IF(event_param.key = "session_id", event_param.value.string_value, NULL )) AS session_id
FROM advanced.app_logs
CROSS JOIN UNNEST(event_params) AS event_param
WHERE event_date BETWEEN "2022-08-01" AND "2022-08-18"
GROUP BY ALL
)
--event_name + screen (필요한 이벤트만 조건 걸어서 사용)
,filter_event_and_concat_event_and_screen AS(
SELECT * EXCEPT(event_name, firebase_screen, event_timestamp),
CONCAT(event_name, "-", firebase_screen) AS event_name_with_screen,
DATETIME(TIMESTAMP_MICROS(event_timestamp), 'Asia/Seoul') AS event_datetime
FROM BASE
WHERE event_name IN ("screen_view","click_payment")
)
--step_number + COUNT
--CASE WHEN 사용
SELECT event_date,
event_name_with_screen,
CASE WHEN event_name_with_screen = "screen_view-welcome" THEN 1
WHEN event_name_with_screen = "screen_view-home" THEN 2
WHEN event_name_with_screen = "screen_view-food_category" THEN 3
WHEN event_name_with_screen = "screen_view-restaurant" THEN 4
WHEN event_name_with_screen = "screen_view-cart" THEN 5
WHEN event_name_with_screen = "click_payment-cart" THEN 6
ELSE NULL
END AS step_number,
COUNT(DISTINCT user_pseudo_id) AS cnt
FROM filter_event_and_concat_event_and_screen
GROUP BY ALL
HAVING step_number IS NOT NULL
ORDER BY event_date
-- food_detail, search, search_result도 파악