연습 문제 (1) -- 1) SELECT user_id , visit_month , LEAD(visit_month, 1) OVER (PARTITION BY user_id ORDER BY visit_month) AS lead_visit_month , LEAD(visit_month, 2) OVER (PARTITION BY user_id ORDER BY visit_month) AS lead2_visit_month FROM advanced.analytics_function_01 ORDER BY user_id -- 2) SELECT user_id , visit_month , LEAD(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month , LEAD(visit_month, 2) OVER (PARTITION BY user_id ORDER BY visit_month) AS after2_visit_month , LAG(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS before_visit_month FROM advanced.analytics_function_01 ORDER BY user_id, visit_month -- 3) 유저가 접속했을 때, 다음 접속까지의 간격을 구하시오. SELECT * , after_visit_month - visit_month AS diff_month FROM ( SELECT user_id , visit_month , LEAD(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month FROM advanced.analytics_function_01 ) ORDER BY user_id , visit_month -- 4) 이 데이터셋을 기준으로 user_id의 첫번째 방문 월, 마지막 방문 월을 구하는 쿼리를 작성해주세요. SELECT * , FIRST_VALUE(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS first_month , LAST_VALUE(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS last_month FROM advanced.analytics_function_01 ORDER BY user_id , visit_month -- 윈도우 함수 - 데이터 범위 지정 SELECT * , SUM(amount) OVER () AS amount_sum , SUM(amount) OVER (ORDER BY order_id) AS cumulative_sum , SUM(amount) OVER (PARTITION BY user_id ORDER BY order_id) AS cumulative_sum_by_user , AVG(amount) OVER (ORDER BY order_id ROWS BETWEEN 5 PRECEDING AND 1 PRECEDING) AS last_5_orders_avg_amount FROM advanced.orders ORDER BY order_id 연습 문제 (2) -- 1) SELECT * , COUNT(user) OVER (PARTITION BY user) AS total_query_cnt FROM advanced.query_logs ORDER BY user -- 2) WITH query_cnt_by_team AS ( SELECT EXTRACT(WEEK FROM query_date) AS week_number , team , user , COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL ) SELECT * , RANK() OVER (PARTITION BY week_number, team ORDER BY query_cnt DESC) AS rk FROM query_cnt_by_team QUALIFY rk = 1 ORDER BY week_number , team , query_cnt DESC -- 3) WITH query_cnt_by_team AS ( SELECT EXTRACT(WEEK FROM query_date) AS week_number , team , user , COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL ) SELECT * , LAG(query_cnt, 1) OVER (PARTITION BY user ORDER BY week_number) AS prev_week_query_cnt FROM query_cnt_by_team ORDER BY user , week_number -- 4) SELECT query_date , team , user , query_cnt , SUM(query_cnt) OVER (PARTITION BY user ORDER BY query_date ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS cumulative_sum FROM ( SELECT query_date , team , user , COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL ) ORDER BY user , query_date -- 5) WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ) SELECT * , LAST_VALUE(number_of_orders IGNORE NULLS) OVER (ORDER BY date) AS before_number_of_orders FROM raw_data -- 6) WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ), filled_data AS ( SELECT * EXCEPT(number_of_orders) , LAST_VALUE(number_of_orders IGNORE NULLS) OVER (ORDER BY date) AS number_of_orders FROM raw_data ) SELECT * , AVG(number_of_orders) OVER (ORDER BY date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS moving_avg FROM filled_data -- 7) WITH step1 AS ( SELECT * EXCEPT(event_params, platform) , DATETIME(TIMESTAMP_MICROS(event_timestamp), "Asia/Seoul") AS event_datetime FROM advanced.app_logs WHERE user_pseudo_id = "1997494153.8491999091" AND event_date = '2022-08-18' ), step2 AS ( SELECT * , DATETIME_DIFF(event_datetime, prev_event_datetime, SECOND) AS second_diff FROM ( SELECT * , LAG(event_datetime) OVER (ORDER BY event_datetime) AS prev_event_datetime FROM step1 ) ORDER BY event_datetime ) SELECT * , SUM(session_start) OVER (ORDER BY event_datetime) AS session_num FROM ( SELECT * , CASE WHEN second_diff IS NULL THEN 1 WHEN second_diff >= 20 THEN 1 ELSE 0 END AS session_start FROM step2 ) ORDER BY event_datetime
연습 문제 (1) 윈도우 함수 -- 1) user들의 다음 접속 월과 다다음 접속 월을 구하는 쿼리를 작성해주세요. SELECT user_id, visit_month, LEAD(visit_month, 1) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month_1, LEAD(visit_month, 2) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month_2, FROM `advanced.analytics_function_01` ORDER BY user_id LEAD 함수와 적절한 수를 이용해 작성 LEAD 함수에 들어가는 인자에 따라 다음인지 다다음인지 정할 수 있음 따로 정해주지 않을 경우 1로 인식 OVER 의 뒷 내용이 생각보다 조금 복잡해 한번에 떠오르지는 않았음 OVER 의 ORDER BY 의 기본은 오름차순 -- 2) user들의 다음 접속 월과 다다음 접속 월, 이전 접속 월을 구하는 쿼리를 작성해주세요. SELECT user_id, visit_month, LEAD(visit_month, 1) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month_1, LEAD(visit_month, 2) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month_2, LAG(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS previous_visit_month_1, FROM `advanced.analytics_function_01` ORDER BY user_id LAG 사용해 이전 값 구해봄 LAG 안에 특별한 숫자를 주지않으니 1로 인식하는 것을 확인함 LEAD 값이 NULL → 해당 값이 마지막 값 LAG 값이 NULL → 해당 값이 첫번째 값 -- 3) user의 다음 접속까지의 간격을 구하시오 SELECT user_id, visit_month, LEAD(visit_month, 1) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month_1, (LEAD(visit_month, 1) OVER (PARTITION BY user_id ORDER BY visit_month) - visit_month) AS diff_month, FROM `advanced.analytics_function_01` ORDER BY user_id SELECT 절에서 만들어진 컬럼은 그대로 사용할 수는 없음 하지만 위와 같이 할 경우 너무 길어지고 복잡해짐(중복됨) → 서브쿼리 사용 SELECT *, (after_visit_month_1 - visit_month) AS diff_month FROM ( SELECT user_id, visit_month, LEAD(visit_month, 1) OVER (PARTITION BY user_id ORDER BY visit_month) AS after_visit_month_1, FROM `advanced.analytics_function_01` ORDER BY user_id ) 쿼리를 최대한 덜 수정하는 방향으로 작성해 볼 것 (2) QUALIFY -- amount_total : 전체 SUM -- cumulative_sum : row 시점에 누적 SUM -- cumulative_sum_by_user : row 시점에 유저별 누적 SUM -- last_5_orders_avg_amount : order_id 기준으로 정렬하고, 직전 5개 주문의 평균 amount SELECT *, SUM(amount) OVER() AS amount_total, SUM(amount) OVER(ORDER BY order_id) AS cumulative_sum, SUM(amount) OVER(PARTITION BY user_id ORDER BY order_id) AS cumulative_sum_by_user, AVG(amount) OVER(ORDER BY order_id ROWS BETWEEN 5 PRECEDING AND 1 PRECEDING) AS last_5_orders_avg_amount FROM advanced.orders ORDER BY order_id BETWEEN 앞에 ROWS 를 빼먹어서 계속 오류를 냈음 -- 1) 사용자별 쿼리를 실행한 총 횟수를 구하는 쿼리를 작성해주세요. -- 단, GROUP BY를 사용해서 집계하는 것이 아닌 query_logs의 데이터의 우측에 새로운 컬럼을 만들어주세요. SELECT *, COUNT(query_date) OVER(PARTITION BY user) AS cnt_by_user FROM advanced.query_logs 데이터에 NULL 값이 없으므로 어떤 열을 세던지 상관 없음 -- 2) 주차별로 팀 내에서 쿼리를 많이 실행한 수를 구한 후, 실행한 수를 활용해 랭킹을 구해주세요. -- 단, 랭킹이 1등인 사람만 결과가 보이도록 해주세요 SELECT *, RANK() OVER(PARTITION BY team, week_key ORDER BY cnt_by_user DESC) AS rnk FROM ( SELECT user, team, IF(query_date < '2024-05-01', 1, 2) AS week_key, COUNT(user) AS cnt_by_user FROM advanced.query_logs GROUP BY ALL ) QUALIFY rnk = 1 ORDER BY team, week_key '주차별’에 대한 아이디어가 잘 떠오르지 않았음 → 데이터의 범위가 좁기 때문에 일단은 IF를 통해 주차를 구분해줌 → 날짜 범위가 넓어지면 어떻게 할지 아직은 모르겠음 서브 쿼리 사용해봄 PARTITION 이 2개임(주차별, 팀별) QUALIFY 사용할 것(생각 못하고 LIMIT 쓰려다 막힘) -- 강의 코드 WITH query_cnt_by_team AS( SELECT EXTRACT(WEEK FROM query_date) AS week_number, team, user, COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL ) SELECT *, RANK() OVER(PARTITION BY week_number, team ORDER BY query_cnt DESC) AS rk FROM query_cnt_by_team QUALIFY rk = 1 ORDER BY week_number, team, query_cnt EXTRACT 함수 통해 ‘주차’ 추출 GROUP BY 후 윈도우 함수 사용 → 유연하게 사용할 것 -- 3) (2번 문제에서 사용한 주차별 쿼리 사용) 쿼리를 실행한 시점 기준 1주 전에 쿼리 실행 수를 별도의 컬럼으로 확인할 수 있는 쿼리를 작성해주세요 SELECT *, LAG(query_cnt) OVER(PARTITION BY user ORDER BY week_number) AS previous_week_cnt FROM query_cnt_by_team ORDER BY user WITH AS 로 만든 테이블 그대로 사용 -- 4) 시간의 흐름에 따라, 일자별로 유저가 실행한 누적 쿼리 수를 작성해주세요 SELECT *, SUM(query_cnt) OVER(PARTITION BY user ORDER BY query_date ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS cumul_cnt FROM ( SELECT user, team, query_date, COUNT(*) AS query_cnt FROM advanced.query_logs GROUP BY ALL ) ORDER BY user, query_date GROUP BY 한 서브쿼리 사용 FRAME 사용이 그렇게 까다롭진 않았음 FRAME 의 defalut 값 → UNBOUNDED PRECEDING ~ CURRENT ROW -- 5) 다음 데이터는 주문 횟수를 나타낸 데이터입니다. 만약 주문 횟수가 없으면 NULL로 기록됩니다. -- 이런 데이터에서 NULL 값이라고 되어있는 부분을 바로 이전 날짜의 값으로 채워주는 쿼리를 작성해주세요 SELECT date, IF(number_of_orders IS NULL, LAG(raw_data.number_of_orders, 1) OVER(ORDER BY date), number_of_orders) AS number_of_orders FROM raw_data 조건문 사용해서 IS NULL 인 값들만 LAG 사용 기존의 number_of_orders가 사라지는 문제가 있음 마지막 날짜는 안채워짐(연속으로 NULL 이라) -- 강의 코드 SELECT *, LAST_VALUE(number_of_orders IGNORE NULLS) OVER(ORDER BY date) AS last_value_orders FROM raw_data LAST_VALUE + IGNORE NULLS 사용 -- 6) 5번 문제에서 NULL을 채운 후, 2일 전 ~ 현재 데이터의 평균을 구하는 쿼리를 작성해주세요(이동 평균) SELECT *, AVG(last_value_orders) OVER(ORDER BY date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS moving_average FROM ( SELECT *, LAST_VALUE(number_of_orders IGNORE NULLS) OVER(ORDER BY date) AS last_value_orders FROM raw_data 서브쿼리 사용했지만 WITH 로 정의해도 됨 → 문제에서는 WITH 가 연속 두번 나오는데, WITH 는 한번만 작성해도 됨(, 로 구분) -- 7) app_logs 테이블에서 Custom Session을 만들어 주세요. 이전 이벤트 로그와 20초가 지나면 새로운 Session을 만들어 주세요. -- Session은 숫자로 (1, 2, 3 ...) 표시해도 됩니다 WITH base AS( SELECT event_date, DATETIME(TIMESTAMP_MICROS(event_timestamp), "Asia/Seoul") AS event_datetime, event_name, user_id, user_pseudo_id, FROM advanced.app_logs WHERE (event_date = "2022-08-18") AND (user_pseudo_id = "1997494153.8491999091") ORDER BY event_timestamp ) SELECT *, SUM(diff_classification) OVER(ORDER BY event_datetime) + 1 AS session_id FROM ( SELECT *, IF(DATETIME_DIFF(event_datetime, before_datetime, second) > 20, 1, 0) AS diff_classification FROM ( SELECT *, LAG(event_datetime) OVER(PARTITION BY user_pseudo_id ORDER BY event_datetime) AS before_datetime, FROM base ) ) ORDER BY event_datetime timestamp와 1초의 관계 : timestamp 1당 1초 아닌가? → 그렇다면 굳이 datetime으로 바꿀 필요가 있나 서브쿼리 2번 중첩해서 사용 IF 문 사용해 20초 차이남 → 1 차이 안남 → 0 으로 파생해 누적합 + 1로 session_id 도출 첫 행 before_datetime에 예외처리 해줄 것 그런데 PARTITION BY 를 매 OVER 안에 무조건 써야 하나? EDA(2) 요일별 접속자 수 + 기간 내 이벤트날 요일별 접속자 평균을 내보자 일 > 토 > 수 > 금 > 목 > 화 > 월 → 역시 쉬는날이 더 배달 수요가 많은건가? → 수요일은 왜 일까? 2022-08-01(월) ~ 2023-01-20(금) 의 데이터 주말(토, 일)이 아닌 공휴일 목록 → 네이버 캘린더 참조 2022-08-15 월 : 광복절 2022-09-09 금 : 추석연휴 2022-09-12 월 : 추석연휴 2022-10-03 월 : 개천절 2023-01-23 월 : 설 연휴 2023-01-24 화 : 설 연휴(대체공휴일) 주요 이벤트 → 특정 사건이 있을 경우 배달 수요가 늘거나 줄지 않을까? / 위키 사이트 참조 2022-08-02 화 : 코로나19 누적 감염자 2천만 명 돌파 2022-08-08 월 : 수도권 기록적인 폭우 및 홍수 2022-10-31 월 : 할로윈데이 / 2022-10-29 토에 이태원 압사 사고 발생 2022-11-08 화 : 한국시리즈 2022-11-17 목 : 2023학년도 대학수학능력시험 2022-11-24 목 : 카타르 월드컵 vs 우루과이 2022-11-28 월 : 카타르 월드컵 vs 가나 2022-12-03 토 : 카타르 월드컵 vs 포르투갈 2022-12-19 월 : 카타르 월드컵 결승 2022-12-25 일 : 크리스마스 등등… “특정 사건”의 중요도를 정성적으로 평가하기에는 어렵다고 느낌 → 오히려 반대로 갑자기 이용자 수의 변화가 급격하게 나타나는 날짜를 위주로 봐야하나? → 그런데 앱 이용자 수 성장 시기에는 항상 상승만해서 보기 애매할 듯 → 요일별 경향을 따져야 할 수도 일요일 < 월요일(개천절) < 화요일 → 앱 이용자 수 상승 예시 월드컵에 따른 이용자 수 변화는 뚜렷하지 않음 수요일에 딱히 뭔가 보이진 않음 → 평일의 절반이 지남에 따른 보상 심리가 원인일수도? 시간대에 따른 이용자 수 마찬가지로 시간대별 평균 이용자 수 구해봄 저녁 시간대(19시 ~ 22시)가 가장 이용자 수 많음 점심 시간대(12시 ~ 14시)가 그 다음 당연하게도 식사 시간에 배달 앱 수요가 많음 시간대 별로 food_id 에 따른 수요가 다른지 확인해볼 것 → 시간대 별 이용자에게 추천해주는 음식 다르게 설정할 수 있음 하루의 기준을 0시로 잡아도 되나?에 대한 의문이 생김(야식 수요) → 이거에 대한 기준을 다시 잡고 일별 이용자 수 다시 구해야 할수도
🔐 이번주차 중요 키워드 : 윈도우 함수, FRAME, QUALIFY ✅ 윈도우 함수 연습문제 1번 1) 사용자별 쿼리를 실행한 총 횟수를 구하는 쿼리 작성 단, group by를 사용해서 집계하는 것이 아닌 query_logs의 데이터 우측에 새로운 컬럼을 만들어주세요. select *, count(query_date) over (partition by user) as query_cnt from advanced.query_logs order by query_cnt desc ✅ 윈도우 함수 연습문제 2번 2) 주차별로 팀 내에서 쿼리를 많이 실행한 수 구하기 2-1) 실행한 수를 활용해 랭킹 구하기 -- 단, 랭킹이 1등인 사람만 결과가 보이도록 해주세요. week_number | team | user | query_cnt | team_rank select * , rank() over(partition by week_number, team order by query_cnt) as team_rank from ( select EXTRACT(week FROM query_date) as week_number, team, user, count(query_date) as query_cnt from advanced.query_logs group by all ) qualify team_rank = 1 order by week_number, team, query_cnt DESC ❗ 새롭게 알게된 함수 : EXTRACT(week FROM query_date) as week_number → 기존의 알고 있던 함수와 같은 결과값 : DATE_TRUNC(query_date, WEEK) AS week_number ❗ qualify team_rank = 1 : QUALIFY 덕분에 서브쿼리 없이 바로 조건에 사용가능함! ✅ 윈도우 함수 연습문제 3번 3) (2번 문제에서 사용한 주차별 쿼리 사용) 쿼리를 실행한 시점 기준 1주전에 쿼리 실행 수를 별도의 컬럼으로 확인할 수 있는 쿼리 작성 select * , LAG(query_cnt, 1) over (partition by user order by week_number) as pre_week_query_cnt from ( select EXTRACT(week FROM query_date) as week_number, team, user, count(query_date) as query_cnt from advanced.query_logs group by all ) ✅ 윈도우 함수 연습문제 4번 4) 시간의 흐름에 따라, 일자별로 유저가 실행한 누적 쿼리수 작성 ## FRAME의 default값은 UNBOUNDED PRECEDING AND CURRENT ROW with query_cnt_by_team as ( select EXTRACT(week FROM query_date) as week_number, team, user, query_date, count(query_date) as query_cnt from advanced.query_logs group by all ) select *, sum(query_cnt) over (partition by user order by query_date ASC) as cumulative_SUM1, sum(query_cnt) over (partition by user order by query_date ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) as cumulative_SUM2, from query_cnt_by_team ❗ FRAME의 default값은 UNBOUNDED PRECEDING AND CURRENT ROW ✅ 윈도우 함수 연습문제 5번 ) 주무횟수 데이터에서 주문횟수가 없으면 NULL로 기록됨. 이런 데이터에서 NULL값이라고 되어있는 부분을 바로 이전 날짜의 값으로 채워주는 쿼리 WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ) select * , ifnull(number_of_orders, LAST_VALUE(number_of_orders IGNORE NULLS) over(order by date)) as filled_orders from raw_data -- 조건절 ifnull 사용할 수 있음. -- LAG()를 사용하면 마지막 NULL값인 경우 채우는 값도 NULL! -- 그렇기 떄문에, LAST_VALUE()인데, NULL은 무시하라는 IGNORE NULLS! ❗️LAST_VALUE에서 IGNORE NULLS 안하면 값은 NULL ✅ 윈도우 함수 연습문제 6번 6) NULL을 채운후, 2일전 ~ 현재 데이터의 평균 (이동평균) WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ) select * , AVG(filled_orders) over (order by date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS avg_orders from ( select * , ifnull(number_of_orders, LAST_VALUE(number_of_orders IGNORE NULLS) over(order by date)) as filled_orders from raw_data ) ✅ 윈도우 함수 연습문제 7번 7) app_logs 테이블에서 custom_session을 만들어 주세요:) 이전 이벤트 로그와 20초가 지나면 새로운 session을 만들어 주세요. event_date | event_timestamp | event_datetime | evnet_name | user_id | user_pseudo_id | before_event_datetime | second_diff | session_start | session_id with base as ( select event_date, DATETIME(TIMESTAMP_MICROS(event_timestamp), 'Asia/Seoul') as event_time, event_name, user_id, user_pseudo_id, from advanced.app_logs where event_date = "2022-08-18" and user_pseudo_id = "1997494153.8491999091" order by event_timestamp ) select *, sum(session_start) over (partition by user_pseudo_id order by event_time) as session_num from ( select *, CASE WHEN before_event_datetime IS NULL THEN 1 WHEN second_diff >=20 THEN 1 ELSE 0 END AS session_start from ( select *, DATETIME_DIFF(event_time, before_event_datetime, SECOND) AS second_diff from ( select *, LAG(event_time,1) over (partition by user_pseudo_id order by event_time) as before_event_datetime from base order by event_time ) ) ) ❗ 새롭게 알게 된 함수 DATETIME_DIFF : 처음에 날짜-시간 차이를 단순히 (-)로만 생각했다가 잘못된 결과 도출 → DATETIME_DIFF 함수로 정답도출!
안녕하세요 카일님, 6-5. 데이터 결과 검증 예시 강의 중 아래 쿼리에서 total_battle_count 와 win_ratio 를 각각 구할때 COUNT(battle_id) 와 COUNT(DISTINCT battle_id) 로 쓰신 이유가 궁금합니다. 무엇이 됐건 둘은 통일되어야 하는게 아닌지, total_battle_count 를 구할때도 엄밀하게는 DISTINCT 를 썼어야 하는게 아닌지 궁금합니다 ! SELECT trainer_id, COUNTIF(battle_result = 'Win') AS win_count, COUNT(battle_id) AS total_battle_count, COUNTIF(battle_result = 'Win') / COUNT(DISTINCT battle_id) AS win_ratio FROM battle_with_result GROUP BY trainer_id
강의 문제 1) user들의 다음 접속월과 다다음 접속월을 구하는 쿼리를 작성해주세요. -- 출제의도: 윈도우 함수(LEAD)를 사용하여 파티션을 나눠 데이터를 탐색할 수 있는가? SELECT user_id , visit_month AS visit_month_m0 , LEAD(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS visit_month_m1 , LEAD(visit_month, 2) OVER (PARTITION BY user_id ORDER BY visit_month) AS visit_month_m2 FROM advanced.analytics_function_01 ORDER BY user_id; 강의 문제 2) user들의 다음 접속월과 다다음 접속월, 이전 접속월을 구하는 쿼리를 작성해주세요 -- 출제의도: 윈도우 함수(LEAD)와 (LAG)을 함께 사용하여 파티션을 나눠 데이터를 탐색할 수 있는가? SELECT user_id , visit_month AS visit_month_m0 , LEAD(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS visit_month_m1 , LEAD(visit_month, 2) OVER (PARTITION BY user_id ORDER BY visit_month) AS visit_month_m2 , LAG(visit_month) OVER (PARTITION BY user_id ORDER BY visit_month) AS visit_month_p1 FROM advanced.analytics_function_01 ORDER BY user_id; 강의 문제 3) Frame 설정을 활용한 윈도우 함수 사용 SELECT order_id , order_date , user_id , amount , SUM(amount) OVER (ORDER BY order_date, order_id ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS amount_total , SUM(amount) OVER (ORDER BY order_date, order_id ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS cumulative_sum , SUM(amount) OVER (PARTITION BY user_id ORDER BY order_date, order_id ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS cumulative_sum_by_user , AVG(amount) OVER (ORDER BY order_date, order_id ROWS BETWEEN 5 PRECEDING AND 1 PRECEDING) AS last_5_orders_avg_amount FROM advanced.orders --QUALIFY last_5_orders_avg_amount >= 150 ORDER BY order_id; 윈도우 함수 연습문제 1) 사용자별 쿼리를 실행한 총 횟수를 구하는 쿼리를 작성해주세요. 단, GROUP BY를 사용해서 집계하는 것이 아닌, query_logs의 데이터 우측에 새로운 컬럼을 만들어주세요. -- 출제의도: 윈도우 함수의 집계 함수 중 COUNT를 사용할 수 있는가? SELECT * , COUNT(user) OVER (PARTITION BY user) AS total_query_cnt FROM advanced.query_logs ORDER BY query_date, user; 윈도우 함수 연습문제 2) 주차별로 팀 내에서 쿼리를 많이 실행한 수를 구한 후, 실행한 수를 활용해 랭킹을 구해주세요. 단, 랭킹이 1등인 사람만 결과가 보이도록 해주세요. -- 출제의도: GROUP BY와 윈도우 함수(순위)를 함께 사용할 수 있는가? SELECT * , RANK() OVER (PARTITION BY week_number, team ORDER BY query_cnt DESC) AS team_rank FROM (SELECT EXTRACT(WEEK FROM query_date) AS week_number , team , user , COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL) QUALIFY team_rank = 1 ORDER BY week_number, team; 윈도우 함수 연습문제 3) (2번 문제에서 사용한 주차별 쿼리 사용) 쿼리를 실행한 시점 기준, 1주 전에 쿼리 실행 수를 별도의 컬럼으로 확인할 수 있는 쿼리를 작성해주세요. -- 출제의도: GROUP BY와 윈도우 함수(LAG)를 함께 사용할 수 있는가? SELECT * , LAG(query_cnt) OVER (PARTITION BY user ORDER BY week_number) AS prev_week_query_count FROM (SELECT user , team , EXTRACT(WEEK FROM query_date) AS week_number , COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL) ORDER BY user, week_number; 윈도우 함수 연습문제 4) 시간 흐름에 따라, 일자별로 유저가 실행한 누적 쿼리 수를 작성해주세요. -- *FRAME의 DEFAULT 값: UNBOUNDED PRECEDING ~ CURRENT ROW -- 출제의도: GROUP BY와 윈도우 함수(SUM-누계합)를 함께 사용할 수 있는가? SELECT * , SUM(query_cnt) OVER (PARTITION BY user ORDER BY query_date) AS cumulative_query_count FROM (SELECT user , team , query_date , COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL) ORDER BY user, query_date; 윈도우 함수 연습문제 5) 다음 데이터는 주문 횟수를 나타낸 데이터입니다. 만약 주문 횟수가 없으면 NULL로 기록됩니다. 이런 데이터에서 NULL 값이라고 되어있는 부분을 바로 이전 날짜의 값으로 채워주는 쿼리를 작성해주세요. -- 출제의도: 윈도우 함수(LAST_VALUE)에서 IGNORE NULLS가 필요한 상황을 이해할 수 있는가? WITH raw_data AS( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ) -- SELECT -- date -- , IFNULL(number_of_orders, LAG(number_of_orders) OVER (ORDER BY date)) AS number_of_orders -- FROM raw_data -- ORDER BY date; SELECT * , LAST_VALUE(number_of_orders IGNORE NULLS) OVER (ORDER BY date) AS number_of_orders FROM raw_data ORDER BY date; 윈도우 함수 연습문제 6) 5번 문제에서 NULL을 채운 후, 2일 전 ~ 현재 데이터의 평균을 구하는 쿼리를 작성해주세요. -- 출제의도: 예외 값을 처리한 이후, 윈도우 함수로 이동 평균을 계산할 수 있는가? WITH raw_data AS( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ) SELECT * , ROUND(AVG(number_of_orders) OVER (ORDER BY date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) , 1) AS moving_avg FROM( SELECT date , IFNULL(number_of_orders, LAG(number_of_orders) OVER (ORDER BY date)) AS number_of_orders FROM raw_data) ORDER BY date; 윈도우 함수 연습문제 7) app_logs 테이블에서 Custom Session을 만들어주세요. 이전 이벤트 로그와 20초가 지나면 새로운 Session을 만들어 주세요. -- *Session은 숫자로(1,2,3…) 표시해도 됩니다. -- **2022-08-18일의 user_pseudo_id(1997494153. 8491999091)은 session_id가 4까지 나옵니다 -- 출제의도: 윈도우 함수를 웹 로그 데이터에 적용하여 활용할 수 있는가? -- Step 1. Session 정보 추출 WITH base AS( SELECT event_date , event_timestamp , DATETIME(TIMESTAMP_MICROS(event_timestamp), 'Asia/Seoul') AS event_datetime , event_name , user_id , user_pseudo_id , DATETIME(TIMESTAMP_MICROS(LAG(event_timestamp) OVER (PARTITION BY user_pseudo_id ORDER BY event_timestamp)), 'Asia/Seoul') AS before_event_datetime FROM advanced.app_logs WHERE event_date = '2022-08-18' AND user_pseudo_id = '1997494153.8491999091' ), -- Step 2. 세션 유지 시간 및 신규 세션 여부 계산 session_info AS( SELECT * , TIMESTAMP_DIFF(event_datetime, before_event_datetime, SECOND) AS second_diff , CASE WHEN TIMESTAMP_DIFF(event_datetime, before_event_datetime, SECOND) >= 20 OR TIMESTAMP_DIFF(event_datetime, before_event_datetime, SECOND) IS NULL THEN 1 ELSE NULL END AS session_start FROM base ) -- Step 3. 신규 세션 id 세팅 SELECT * , SUM(session_start) OVER (PARTITION BY user_pseudo_id ORDER BY event_datetime) AS session_id FROM session_info ORDER BY event_date, event_timestamp;
안녕하세요 카일님, JOIN 연습문제 4번에서 작성 순서 및 효율성 관련해서 질문드리고 싶습니다. 카일님의 풀이에서는 Active, Training 의 Status 를 필터링 후 pokemon, trainer 테이블을 LEFT JOIN 하시고 마지막에 'Master' 를 필터링하셨는데, 저 같은 경우에는 각각의 테이블마다 필요한 부분을 먼저 필터링 후에 마지막에는 COUNT 만 할 수 있게 진행하였습니다. Trainer 테이블에서는 'Master' 필터 Trainer_pokemon 테이블에서는 'Active', 'Training' 필터 결과적으로 결과값은 올바르게 나오는데 카일님께서 앞선 강의에서 필터링을 먼저 하는 것이 효율적이라 하셨던 것 같아서 질문드리게 되었습니다. 감사합니다. SELECT p.type1, COUNT(tp.id) AS pokemon_cnt -- t.*, -- tp.pokemon_id, -- tp.status, -- p.type1 FROM( SELECT id, achievement_level FROM `basic.trainer` WHERE achievement_level = 'Master' ) AS t LEFT JOIN( SELECT * FROM `basic.trainer_pokemon` WHERE status IN ('Active', 'Training') ) AS tp ON t.id = tp.trainer_id LEFT JOIN `basic.pokemon` AS p ON tp.pokemon_id = p.id WHERE tp.pokemon_id IS NOT NULL GROUP BY p.type1 ORDER BY 2 DESC LIMIT 1;
윈도우 탐색 함수 연습문제 (1) 연습문제 1 -- 문제 1) USER의 다음 접속월, 다다음 접속 월 SELECT user_id, visit_month, LEAD(visit_month,1) OVER(PARTITION BY user_id ORDER BY visit_month) AS next_month, LEAD(visit_month,2) OVER(PARTITION BY user_id ORDER BY visit_month) AS the_month_after_next FROM `avdanced.analytics_function_01` (2) 연습문제 2 -- 문제 2) USER의 다음 접속월, 다다음 접속 월, 이전 접속 월 SELECT user_id, visit_month, LEAD(visit_month,1) OVER(PARTITION BY user_id ORDER BY visit_month) AS next_month, LEAD(visit_month,2) OVER(PARTITION BY user_id ORDER BY visit_month) AS the_month_after_next, LAG(visit_month,1) OVER(PARTITION BY user_id ORDER BY visit_month) AS last_month FROM `avdanced.analytics_function_01` 윈도우 함수 FRAME 연습문제 연습문제 (1~4) SELECT -- 1)모든 주문량 SUM(amount) OVER() AS amount_total, -- 2)특정주문시점에서 누적주문량 #SUM(amount) OVER(partition by order_date) AS cumulative_sum, SUM(amount) OVER (ORDER BY order_date) AS cumulative_sum, -- 3)고객별 주문 시점에서 누적 주문량 #SUM(amount) OVER(partition by user_id) AS cumulative_sum_by_user, SUM(amount) OVER(partition by user_id ORDER BY order_id) AS cumulative_sum_by_user, -- 4) 최근 직전 5개 평균 주문량 AVG(amount) OVER(ROWS BETWEEN 5 PRECEDING AND 1 PRECEDING) AS last_5_orders_avg_amount, AVG(amount) OVER(ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING) AS last_5_unbounded_orders_avg_amount, AVG(amount) OVER(ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS all_orders_avg_amount FROM `avdanced.orders` 윈도우 함수 (1) 연습문제 1 -- 연습문제1) 사용자별 쿼리 실행 횟수 WITH base AS( SELECT user, team, query_date, COUNT(*) OVER(PARTITION BY user) AS total_query_cnt, FROM `avdanced.query_logs` ) SELECT * FROM base (2) 연습문제 2 -- 연습문제2) 주차별 팀내 쿼리 실행한 수 (RANK 1만 보이도록) WITH base2 AS( SELECT EXTRACT(WEEK FROM query_date) AS week_number, team, user, COUNT(*) OVER(PARTITION BY EXTRACT(WEEK FROM query_date) ,user ORDER BY EXTRACT(WEEK FROM query_date) ) AS query_cnt, FROM `avdanced.query_logs` ORDER BY EXTRACT(WEEK FROM query_date) ) SELECT DISTINCT *, RANK() OVER(PARTITION BY team,week_number ORDER BY total_query_cnt DESC) AS team_rank FROM base2 QUALIFY team_rank = 1 ORDER BY week_number, team 강의자료의 코드 -- 2) 주차별로 팀 내에서 쿼리를 많이 실행한 수를 구한 후, 실행한 수를 활용해 랭킹을 구해주세요. 단, 랭킹이 1등인 사람만 결과가 보이도록 해주세요 -- 주차별로 개인당 실행한 쿼리 횟수 -- 위 쿼리 횟수를 기반으로 랭킹 -- 랭킹을 기반으로 필터링(랭킹=1) -- 문제의 의도 : 원본 데이터 => 1 row마다 데이터가 있고, 그걸 집계해서 사용. GROUP BY => 윈도우 함수 WITH query_cnt_by_team AS ( SELECT EXTRACT(WEEK FROM query_date) AS week_number, team, user, COUNT(user) AS query_cnt FROM advanced.query_logs GROUP BY ALL ) SELECT *, RANK() OVER(PARTITION BY week_number, team ORDER BY query_cnt DESC) AS rk FROM query_cnt_by_team -- QUALIFY : 윈도우 함수의 조건을 설정할 때 사용 QUALIFY rk = 1 ORDER BY week_number, team, query_cnt DESC COUNT의 윈도우 함수 대신에 GROUP BY를 사용하는 풀이도 있다는 것을 알게 되었다! 너무 어렵게 생각하지 말기! (3) 연습문제 3 WITH base2 AS( SELECT EXTRACT(WEEK FROM query_date) AS week_number, team, user, COUNT(*) OVER(PARTITION BY EXTRACT(WEEK FROM query_date) ,user ORDER BY EXTRACT(WEEK FROM query_date) ) AS query_cnt, FROM `avdanced.query_logs` #QUALIFY team_rank = 1 ORDER BY EXTRACT(WEEK FROM query_date) ), base3 AS( SELECT DISTINCT *, RANK() OVER(PARTITION BY team,week_number ORDER BY query_cnt DESC) AS team_rank FROM base2 QUALIFY team_rank = 1 ORDER BY week_number, team ) -- 연습문제3) 쿼리 실행 시점 1주전 쿼리 실행 SELECT DISTINCT *, LAG(query_cnt,1) OVER(PARTITION BY user ORDER BY week_number) AS prev_week_query_count FROM base2 GROUP BY ALL ORDER BY user, week_number (4) 연습문제 4 --연습문제4) SELECT *, SUM(query_count) OVER(PARTITION BY user ORDER BY query_date) AS culmulative_query_count, SUM(query_cnt) OVER(PARTITION BY user ORDER BY query_date ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS cumulative_sum2 FROM( SELECT DISTINCT *, COUNT(user) OVER(PARTITION BY query_date, user) AS query_count, FROM `avdanced.query_logs` ) ORDER BY user,query_date QUALIFY 로 조건설정을 하여 두 값이 같은 지 비교하는 법이 인상깊었던 문제 (QUALIFY cumulative_sum != cumulative_sum2) (5) 연습문제 5 나의 풀이 WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ) --연습문제 5) null에 이전 값 삽입 SELECT raw_data.date, IF(raw_data.number_of_orders IS NULL, LAG(raw_data.number_of_orders,1) OVER(ORDER BY date), raw_data.number_of_orders) FROM raw_data 강의풀이 WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ), -- LAG로 직전 값 가져오면 되지 않을까? -- number_of_orders가 null이면, before_number_of_orders를 가져와라! -- 아래 쿼리는 어려운 방법 -- 그 다음 방법 : LAST VALUE를 쓰자! => 값이 없으면 NULL이 뜬다! -- FIRST_VALUE, LAST_VALUE => NULL을 포함해서 연산 -- 출제 의도 : NULL을 제외해서 연산하고 싶으면 IGNORE NULLS을 쓰면 된다! -- SELECT -- *, -- IF(number_of_orders IS NULL, before_number_of_orders, number_of_orders) AS filled_orders -- -- Number of arguments does not match for function IF. Supported signature: IF(BOOL, ANY, ANY) at [89:3] -- -- False일 때 인자를 추가하지 않아서 생긴 오류 -- FROM ( -- SELECT -- *, -- LAG(number_of_orders) OVER(ORDER BY date) AS before_number_of_orders -- FROM raw_data -- ) filled_data AS ( SELECT * EXCEPT(number_of_orders), LAST_VALUE(number_of_orders IGNORE NULLS) OVER(ORDER BY date) AS number_of_orders FROM raw_data -- Syntax error: Expected keyword DEPTH but got identifier "filled_data" at [104:6] : WITH문을 두개 작성했는데 WITH 쉼표 쓰고 구분! ) LAST_VALUE, FIRST_VALUE를 사용하는 풀이 법에 대하여 알게 됨 (6) 연습문제 6 WITH raw_data AS ( SELECT DATE '2024-05-01' AS date, 15 AS number_of_orders UNION ALL SELECT DATE '2024-05-02', 13 UNION ALL SELECT DATE '2024-05-03', NULL UNION ALL SELECT DATE '2024-05-04', 16 UNION ALL SELECT DATE '2024-05-05', NULL UNION ALL SELECT DATE '2024-05-06', 18 UNION ALL SELECT DATE '2024-05-07', 20 UNION ALL SELECT DATE '2024-05-08', NULL UNION ALL SELECT DATE '2024-05-09', 13 UNION ALL SELECT DATE '2024-05-10', 14 UNION ALL SELECT DATE '2024-05-11', NULL UNION ALL SELECT DATE '2024-05-12', NULL ), null_is_lag AS( --연습문제 5) null에 이전 값 삽입 SELECT raw_data.date, IF(raw_data.number_of_orders IS NULL, LAG(raw_data.number_of_orders,1) OVER(ORDER BY date), raw_data.number_of_orders) AS number_of_orders FROM raw_data ) -- 연습문제 6) 이동평균 SELECT *, AVG(nl.number_of_orders) OVER(ORDER BY nl.date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS moving_avg FROM null_is_lag AS nl FRAME절을 사용할 때, AND를 기준으로 앞에는 뒤의 값보다 반드시 이전 행을 가리키는 구문이 와야 함! (7) 연습문제 7 -- 1. TIMESTAMP → DATETIME -- 2. SECOND_DIFF 생성 : uSER로 묶어서 - -- 3. SESSION_START생성 : USER로 묶어서 LAG(DATA,1)이 NULL이면 1, SECOND_DIFF가 20이상이면 +1 -- 4. SESSION_ID생성: SESSION_START가 1일 경우 SESSION_ID +1, NULL일 경우 LAG(DATA,1) WITH add_date AS ( -- 1. TIMESTAMP → DATETIME SELECT event_date, event_timestamp, DATETIME(TIMESTAMP_MICROS(event_timestamp)) AS event_datetime, event_name, user_id, user_pseudo_id, LAG(DATETIME(TIMESTAMP_MICROS(event_timestamp))) OVER(PARTITION BY user_pseudo_id ORDER BY event_timestamp) AS before_event_datetime FROM `avdanced.app_logs_temp` --,UNNEST(event_params) AS param -- FROM 절 안에서 UNNEST를 사용 WHERE event_date ="2022-08-18" AND user_pseudo_id = "1997494153.8491999091" ), add_diff AS ( -- 2. SECOND_DIFF 생성 : uSER로 묶어서 - SELECT *, DATE_DIFF(event_datetime, before_event_datetime,SECOND) AS second_diff, FROM add_date ), add_session AS( -- 3. SESSION_START생성 : USER로 묶어서 LAG(DATA,1)이 NULL이면 1, SECOND_DIFF가 20이상이면 +1 SELECT *, IF(second_diff IS NULL OR second_diff >=20, 1, NULL) AS session_start FROM add_diff ) -- 4. SESSION_ID생성 *, SUM(session_start) OVER(PARTITION BY user_pseudo_id ORDER BY event_datetime) AS session_num FROM add_session ORDER BY event_datetime user_id와 user_pseudo_id는 다르다. (계정이 존재하면 USER_ID, 없어도 USER_PSEUDO_ID를 통해 활동기록이 남는다.) IF문은 행 단위로 작동, SUM은 특정 파티션에 대한 누적합을 계산함으로 아래 코드가 작동하지 않음 SUM(session_start) OVER(PARTITION BY user_pseudo_id ORDER BY event_datetime) AS session_num CTE를 사용하는 것과 서브쿼리를 사용하는 방법 중 상황에 맞게 적절히 혼용할 수 있다는 것을 알게 됨!
[1~2번]윈도우 함수(탐색 함수) 연습 문제 1번 select *, lead(visit_month, 1) over(partition by user_id order by visit_month) as next_visit_month, lead(visit_month, 2) over(partition by user_id order by visit_month) as two_next_visit_month from `advanced.analytics_function_01 2번 select *, lead(visit_month, 1) over(partition by user_id order by visit_month) as next_visit_month, lead(visit_month, 2) over(partition by user_id order by visit_month) as two_next_visit_month, lag(visit_month, 1) over(partition by user_id order by visit_month) as before_visit_month from `advanced.analytics_function_01` [1~4번] 윈도우 함수 Frame 연습 문제 1번 /* 전체 계산 할 때 왜 over()를 써야 하는지 몰랐습니다. 그러나 일반 집계 함수의 방식과 비교해보니, 사용 이유를 알게 됐습니다. 집계함수를 사용할 때 over()를 사용했다는 건 '윈도우 함수가 적용된 결과값(각 행마다 결과 값이 반환 되는 것)으로 보겠다'는 의미인 걸 알게 됐습니다. 같은 집계함수를 사용하더라도 '어떤 방식(결과를 1개 행으로만 받아볼지, 여러개 값으로 받아볼지)으로 결과 값을 받아 볼 지' 생각해 봐야 한다는 걸 알았습니다. */ select *, sum(amount) over() as amount_total from `advanced.orders` 2번 /* 처음에는 누적합을 구하는 건데 왜 over() 안에 order by를 써야하는지 이해하지 못했습니다. 그러나 누적합도 "어떤 방식으로 계산할지 '계산 방식'"을 지정해 줘야 한다는 걸 알게 됐습니다. /* select *, sum(amount) over(order by order_id) as cumulative_sum from `advanced.orders` 3번 select *, sum(amount) over(partition by user_id order by order_id) as cumulative_sum_by_user from `advanced.orders` 4번 /* over() 안에 사용하는 order by와 over() 밖에서 사용하는 order by를 명확히 이해했습니다. 윈도우 함수에서 사용하는 order by는 '계산 순서'를 정하는 것이고, 윈도우 함수 밖에서 사용한 order by는 '눈에 보이는 정렬'을 컨트롤 한다는 걸 이해했습니다. 이해할 때 '쿼리 작동 순서'를 생각했더니, 큰 도움이 됐습니다. */ select *, avg(amount) over(order by order_id rows between 5 preceding and 1 preceding) as last_5_orders_avg_amount from `advanced.orders` order by order_id [1~7번] 윈도우 함수 연습 문제 1번 select *, count(user) over(partition by user) as total_query_cnt from `advanced.query_logs` order by query_date 2번 /* - date 형식 데이터를 extract()로 주 형식으로 바꾸는 걸 알게 됐습니다 - partition by 내에 2개 이상 컬럼을 넣어도 된다는 걸 알게 됐습니다. */ with user_query_cnt as ( select extract(week from query_date) as week_number, team, user, count(user) as query_cnt from `advanced.query_logs` group by all ) select *, rank() over(partition by week_number, team order by query_cnt desc) as team_rank from user_query_cnt qualify team_rank = 1 order by 1,2 3번 /* - lag()와 함께 쓰는 over() 사용법이 헷갈렸는데, 실행 후 수정하는 방향으로 쿼리를 작성해 보라고 하신 점이 인상 깊게 와닿았습니다. */ with base as ( select user, team, extract(week FROM query_date) as week_number, count(user) as query_count from advanced.query_logs group by all ) select *, lag(query_count, 1) over(partition by user order by week_number) as pre_query_cnt from base 4번 /* qualify cumulative_query_cnt != cumulative_query_cnt2 위 쿼리로 정합성을 체크하는 방식을 배우게 됐습니다. */ ## 4번 with query_cnt as ( select *, count(user) as query_count from advanced.query_logs group by all ) select *, sum(query_count) over(partition by user order by query_date) as cumulative_query_cnt, sum(query_count) over(partition by user order by query_date rows between unbounded preceding and current row) as cumulative_query_cnt2 from query_cnt -- qualify cumulative_query_cnt != cumulative_query_cnt2 5번 /* - 4번에서 배운 정합성 체크하는 쿼리를 이용해서 컬럼 내 값이 있는지 테스트 해봤습니다. */ ## 정합성 테스트 -- select -- * -- from ( -- select -- *, -- last_value(number_of_orders) over(order by date) as last_values_order -- from raw_data -- ) as test1 -- where -- number_of_orders != last_values_order select *, last_value(number_of_orders ignore nulls) over(order by date) as last_values_order from raw_data 6번 /* 처음에는 1. 평균을 구하고 2. 해당 테이블을 서브쿼리로 두고, lag(컬럼, 2)를 해야 하는줄 알았습니다. 그러나 Frame 사용 방법이 생각났고, 조건에 맞게 범위를 설정해서 문제를 해결했습니다. */ ## 6번 , base2 as ( select *, last_value(number_of_orders ignore nulls) over(order by date) as last_values_order, from raw_data ) select *, avg(last_values_order) over(order by date rows between 2 preceding and current row) as moving_avg from base2 7번 /* - 타임스탬프 데이터를 다루는 다양한 방법을 알게 됐습니다. - 실무에서는 세션 기준과 보고 싶은 유저 혹은 이벤트를 지정해서 사용해도 된다는 걸 알게 됐습니다. */ with base as ( select event_date, event_timestamp, datetime(timestamp_micros(event_timestamp), 'Asia/Seoul') as event_datetime, event_name, user_id, user_pseudo_id from `advanced.app_logs` where event_date = '2022-08-18' and user_pseudo_id = '1997494153.8491999091' ) select *, sum(session_start) over(order by event_datetime) as session_id from ( select *, case when second_diff >= 20 then 1 when second_diff is null then 1 else 0 end session_start from ( select *, timestamp_diff(event_datetime, before_event_datetime, second) as second_diff from ( select *, lag(event_datetime, 1) over(partition by user_pseudo_id order by event_datetime) as before_event_datetime from base ) as base2 ) as base3 ) as base4
윈도우 함수 (탐색 함수) 연습 문제 1번 SELECT user_id, visit_month, lead(visit_month, 1) OVER(PARTITION BY user_id ORDER BY visit_month) AS next_1month, lead(visit_month, 2) OVER(PARTITION BY user_id ORDER BY visit_month) AS next_2month, FROM advanced.analytics_function_01 ORDER BY 1,2; 2번 SELECT user_id, visit_month, lead(visit_month, 1) OVER(PARTITION BY user_id ORDER BY visit_month) AS next_1month, lead(visit_month, 2) OVER(PARTITION BY user_id ORDER BY visit_month) AS next_2month, lag(visit_month, 1) OVER(PARTITION BY user_id ORDER BY visit_month) AS pre_1month FROM advanced.analytics_function_01 ORDER BY 1,2; 윈도우 함수 연습 문제 1번 SELECT user, team, query_date, count(user) over(PARTITION BY user) AS total_query_cnt FROM advanced.query_logs ORDER BY query_date, user 2번 WITH base AS ( SELECT EXTRACT(week FROM query_date) AS week_number, team, user, count(user) AS query_cnt FROM advanced.query_logs GROUP BY 1,2,3 ) SELECT *, rank() OVER(PARTITION BY week_number, team ORDER BY query_cnt desc) AS team_rank FROM base QUALIFY team_rank = 1 ORDER BY week_number, query_cnt desc; 3번 WITH base AS ( SELECT EXTRACT(week FROM query_date) AS week_number, team, user, count(user) AS query_cnt FROM advanced.query_logs GROUP BY 1,2,3 ) SELECT *, lag(query_cnt) OVER(PARTITION BY team, user ORDER BY week_number asc) AS prev_week_query_cnt FROM base ORDER BY team, user, week_number; 4번 WITH base AS ( SELECT user, team, query_date, COUNT(user) as query_count FROM advanced.query_logs GROUP BY 1,2,3 ) SELECT user, team, query_date, query_count, SUM(query_count) OVER(PARTITION BY team, user ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS cumulative_query_count FROM base ORDER BY team, user, query_date; 5번 WITH raw_data AS( SELECT DATE'2024-05-01'AS date,15 AS number_of_orders UNION ALL SELECT DATE'2024-05-02',13 UNION ALL SELECT DATE'2024-05-03',NULL UNION ALL SELECT DATE'2024-05-04',16 UNION ALL SELECT DATE'2024-05-05',NULL UNION ALL SELECT DATE'2024-05-06',18 UNION ALL SELECT DATE'2024-05-07',20 UNION ALL SELECT DATE'2024-05-08',NULL UNION ALL SELECT DATE'2024-05-09',13 UNION ALL SELECT DATE'2024-05-10',14 UNION ALL SELECT DATE'2024-05-11',NULL UNION ALL SELECT DATE'2024-05-12',NULL ) SELECT date, IF(number_of_orders is null , last_value(number_of_orders IGNORE NULLS) OVER(ORDER BY date asc), number_of_orders) AS number_of_orders_not_null FROM raw_data; 6번 WITH raw_data AS( SELECT DATE'2024-05-01'AS date,15 AS number_of_orders UNION ALL SELECT DATE'2024-05-02',13 UNION ALL SELECT DATE'2024-05-03',NULL UNION ALL SELECT DATE'2024-05-04',16 UNION ALL SELECT DATE'2024-05-05',NULL UNION ALL SELECT DATE'2024-05-06',18 UNION ALL SELECT DATE'2024-05-07',20 UNION ALL SELECT DATE'2024-05-08',NULL UNION ALL SELECT DATE'2024-05-09',13 UNION ALL SELECT DATE'2024-05-10',14 UNION ALL SELECT DATE'2024-05-11',NULL UNION ALL SELECT DATE'2024-05-12',NULL ), fill_null AS ( SELECT date, IF(number_of_orders is null , last_value(number_of_orders IGNORE NULLS) OVER(ORDER BY date asc), number_of_orders) AS number_of_orders_not_null FROM raw_data ) SELECT date, number_of_orders_not_null, avg(number_of_orders_not_null) OVER(ORDER BY date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS moving_avg FROM fill_null; 7번 WITH base AS ( SELECT event_date, event_timestamp, DATETIME(TIMESTAMP_MICROS(event_timestamp), 'Asia/Seoul') AS event_datetime, event_name, user_id, user_pseudo_id FROM advanced.app_logs WHERE event_date = '2022-08-18' ), diff_date AS ( SELECT *, DATETIME_DIFF(event_datetime, pre_event_time, second) AS date_diff_sec FROM ( SELECT *, LAG(event_datetime, 1) OVER(PARTITION BY user_pseudo_id ORDER BY event_datetime asc) AS pre_event_time FROM base ) ), session_start AS ( SELECT *, CASE WHEN pre_event_time IS NULL THEN 1 WHEN date_diff_sec >= 20 THEN 1 END AS start_session FROM diff_date ) SELECT event_date, event_datetime, event_name, user_id, user_pseudo_id, date_diff_sec, SUM(start_session) OVER(PARTITION BY user_pseudo_id ORDER BY event_datetime) AS session_id FROM session_start ORDER BY user_pseudo_id, event_datetime;
[ARRAY, STRUCT] array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요. SELECT title, #genres, genre FROM `advanced.array_excercises` ae cross join unnest(genres) as genre; 2. array_exercies 테이블에서 각 영화(title)별로 배우(actor)와 배역(character)을 보여주세요 배우와 배역은 별도의 컬럼으로 나와야 합니다. select title, actor.actor, actor.character from `advanced.array_excercises` cross join unnest(actors) as actor; array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르 (genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다. select title, actor.actor, actor.character, genre from `advanced.array_excercises` cross join unnest(actors) as actor cross join unnest(genres) as genre; 앱 로그 데이터(app_logs) 배열 풀기 select user_id, event_date, event_name, user_pseudo_id, params.key, params.value.string_value as string_value, params.value.int_value as int_value from `advanced.app_logs` cross join unnest(event_params) as params where event_date = "2022-08-01 [PIVOT] 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, # amount의 합 sum(if(user_id=1, amount, 0)) as user_1, sum(if(user_id=2, amount, 0)) as user_2, sum(if(user_id=3, amount, 0)) as user_3 FROM `advanced.orders` group by order_date order by order_date; orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요. user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다 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; 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. select user_id, max(if(order_date = "2023-05-01", 1, 0)) as `2023-05-01`, max(if(order_date = "2023-05-02", 1, 0)) as `2023-05-02`, max(if(order_date = "2023-05-03", 1, 0)) as `2023-05-03`, max(if(order_date = "2023-05-04", 1, 0)) as `2023-05-04`, max(if(order_date = "2023-05-05", 1, 0)) as `2023-05-05` from `advanced.orders` group by user_id order by user_id; user_id = 32888이 카트 추가하기(click_cart)를 누를때 어떤 음식(food_id)을 담았나요? with base as ( select -- * EXCEPT(event_params), # * except(컬럼) :컬럼을 제외하고 모두 다 보여줘 -- param 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.string_value, null)) as food_id, 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 from `advanced.app_logs` cross join unnest(event_params) as param group by all ) select user_id, event_date, count(user_id) as user_cnt, food_id from base where user_id = 32888 and event_name = "click_cart" group by all [퍼널분석] with base as ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, platform, max(if(event_param.key = "firebase_screen", event_param.value.string_value, null)) as firebase_screen, 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 ), filter_event 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") ), screen_view as( 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 group by all having step_number is not null order by event_date ) select event_date, max(if(event_name_with_screen = "screen_view-welcome", cnt, null)) as screen_view_welcome, max(if(event_name_with_screen = "screen_view-home", cnt, null)) as screen_view_home, max(if(event_name_with_screen = "screen_view-food_category", cnt, null)) as screen_view_food_category, max(if(event_name_with_screen = "screen_view-restaurant", cnt, null)) as screen_view_restaurant, max(if(event_name_with_screen = "screen_view-cart", cnt, null)) as screen_view_cart, max(if(event_name_with_screen = "click_payment-cart", cnt, null)) as click_payment_cart from screen_view group by all order by event_date
--연습문제 CREATE OR REPLACE TABLE advanced.array_exercises AS SELECT movie_id, title, actors, genres FROM ( SELECT 1 AS movie_id, 'Avengers: Endgame' AS title, ARRAY<STRUCT<actor STRING, character STRING>>[ STRUCT('Robert Downey Jr.', 'Tony Stark'), STRUCT('Chris Evans', 'Steve Rogers') ] AS actors, ARRAY<STRING>['Action', 'Adventure', 'Drama'] AS genres UNION ALL SELECT 2, 'Inception', ARRAY<STRUCT<actor STRING, character STRING>>[ STRUCT('Leonardo DiCaprio', 'Cobb'), STRUCT('Joseph Gordon-Levitt', 'Arthur') ], ARRAY<STRING>['Action', 'Adventure', 'Sci-Fi'] UNION ALL SELECT 3, 'The Dark Knight', ARRAY<STRUCT<actor STRING, character STRING>>[ STRUCT('Christian Bale', 'Bruce Wayne'), STRUCT('Heath Ledger', 'Joker') ], ARRAY<STRING>['Action', 'Crime', 'Drama'] ) -- 1) array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요 select title , genres_new from advanced,.array_exercise AS a, UNNEST(genres) as genres_new -- 2) array_exercises 테이블에서 각 영화(title)별로 배우(actor)와 배역(character)을 보여주세요. 배우와 배역은 별도의 컬럼으로 나와야 합니다 select title , actors_new.actor , actors_new.character from advanced,.array_exercise AS a, UNNEST(actors) as actors_new --3) array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르 (genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다 --방법 1 with gen as ( select title , genres_new from advanced,.array_exercise AS a, UNNEST(genres) as genres_new ) , actors as ( select title , actors_new.actor , actors_new.character from advanced,.array_exercise AS a, UNNEST(actors) as actors_new ) select from gen g join actors a on g.title=a.title --방법 2 select title , actors_new.actor , actors_new.character , genre_new from advanced,.array_exercise AS a, UNNEST(actors) as actors_new, UNNEST(genres) as genre_new --방법 3 select title , actors_new.actor , actors_new.character , genre_new from advanced.array_exercise cross join UNNEST(actors) as actors_new cross join UNNEST(genres) as genre_new where actors_new.actor 로 조건을 걸어야함 --actor(키값바로) 또는 actors_new로는 안된다 actors_new는 스트럭트 구조이고 actor는 이전 값임 --4) 앱 로그 데이터(app_logs)의 배열을 풀어주세요 --하루 사용자 집계, 어떤 이벤트가 있는가? select user_id , event_date , event_name , user_pseudo_id , event_component.key , event_component.value.string_value , event_component.value.int_value from app_logs as app, UNNEST(event_pharams) as event_component where event_date = '2022-08-11' --피봇 과제 --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 with raw as ( select user_id , order_date , sum(amount) as amounts from orders ) SELECT order_date , MAX(IF(user_id=1, amounts, NULL)) AS user_1 , MAX(IF(user_id=2, amounts, NULL)) AS user_2 , MAX(IF(user_id=3, amounts, NULL)) AS user_3 ... FROM raw GROUP BY order_date --2) orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요. user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다 with raw as ( --혹시나 하나의 유저가 하루에 여러 주문을 했을수도 있을것 같아서 이것 사용(MAX 쓸 예정이라서) select user_id , order_date , sum(amount) as amounts from orders ) SELECT user_id , MAX(IF(order_date='2023-05-01', amounts, NULL)) AS '2023-05-01' , MAX(IF(order_date='2023-05-02', amounts, NULL)) AS '2023-05-02' , MAX(IF(order_date='2023-05-03', amounts, NULL)) AS '2023-05-03' ... FROM raw GROUP 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 with raw as ( select user_id , order_date , sum(amount) as amounts , count(distinct order_id) as order_cnt from orders ) SELECT user_id , MAX(IF(order_date='2023-05-01', 1, 0)) AS '2023-05-01' , MAX(IF(order_date='2023-05-02', 1, 0)) AS '2023-05-02' , MAX(IF(order_date='2023-05-03', 1, 0)) AS '2023-05-03' --second case , MAX(IF(order_date='2023-05-01', order_cnt, 0)) AS '2023-05-01' , MAX(IF(order_date='2023-05-02', order_cnt, 0)) AS '2023-05-02' , MAX(IF(order_date='2023-05-03', order_cnt, 0)) AS '2023-05-03' ... FROM raw GROUP BY user_id -- 4)user_id = 32888이 카트 추가하기(click_cart)를 누를때 어떤 음식(food_id)을 담았나요? -- ##데이터 조회할때 유용한 except(column):특정 컬럼 제외하고 모두 다 -- select * except(event_params) -- GROUP BY ALL with raw as ( select user_id , event_date , event_name , user_pseudo_id , event_component.key , event_component.value.int_value as food_id from app_logs as app, UNNEST(event_pharams) as event_component where user_id = 32888 and event_name = 'click_cart' and event_component.key = 'food_id' ) select distinct event_time --나는 일별로 보고 싶어서 추가 , food_id from raw --성윤님 강의 내용 select user_id , event_name , event_date , event_timestamp --같은일에 중복 주문이 있을까봐서 , MAX(IF(event_component.key='food_id', event_component.value.int_value, NULL)) AS food_id from app_logs as app, UNNEST(event_pharams) as event_component where user_id = 32888 and event_name = 'click_cart' and event_component.key = 'food_id' GROUP BY ALL --알아서 컬럼들 픽 --퍼널 별 유저 수 집계 with raw as ( select user_id , event_date , event_timestamp , event_name , user_pseudo_id , platform , event_component.key , event_component.value.string_value , event_component.value.int_value , MAX(IF(event_component.key = "firebase_screen", event_component.value.string_value, NULL)) AS firebase_screen -- , MAX(IF(event_component.key = "food_id", event_component.value.int_value, NULL)) AS food_id , MAX(IF(event_component.key = "session_id", event_component.value.int_value, NULL)) AS session_id from app_logs as app, UNNEST(event_pharams) as event_component where event_date BETWEEN "2022-08-01" AND "2022-08-18" group by all ) , 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") ) 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, step_number 강의 노트 select [0,1,1,2,3,4] as array_practice array<int64>[0,1,3] as array_practice generate_array(1,5,2) generate_date_array('2024-01-01', '2024-02-01', interval 1 week) WITH programming_languages AS ( SELECT "python" AS programming_language UNION ALL SELECT "go" UNION ALL SELECT "scala" ) select array_agg(programming_languages) as output from programming_languages --배열에 접근하기 offset: #0 ordinal: #1 #out of range를 방지하기 위해서 safe_ 추가하기 --사용 예시 select some_numbers[safe_offset(1)] as second_value 컬럼명[safe_offset(가져오고 싶은 위치)] Array(like list): 비슷한 카테고리에 대해 데이터를 저장할때 예시) 메뉴(컬럼): 돼지국밥, 떡볶이, 치킨 Struct(like dict): 다양한 속성에 대해 데이터를 한 컬럼에 다 넣고 싶을때 예시) 주소록(컬럼): 이름, 전화번호,이메일, 생일 등등 SELECT (1,2,3) AS struct_test SELECT STRUCT<hi INT64, hello INT64, awesome STRING>(1, 2, 'HI') AS struct_test SELECT struct_test.hi, struct_test.hello FROM ( SELECT STRUCT<hi INT64, hello INT64, awesome STRING>(1, 2, 'HI') AS struct_test ) -- UNNEST를 사용해 중첩된 데이터 구조 풀기(평면화, Flatten) WITH example_data AS( SELECT 'kyle' AS name, ['Python', 'SQL', 'R', 'Julia', 'Go'] AS preferred_language, 'Incheon' AS hometown UNION ALL SELECT 'max' AS name, ['Python', 'SQL', 'Scala', 'Java', 'Kotlin'] AS preferred_language, 'Seoul' AS hometown UNION ALL SELECT 'yun' AS name, ['Python', 'SQL'] AS preferred_language, 'Incheon' AS hometown ) SELECT name, pref_lang, hometown FROM example_data CROSS JOIN UNNEST(preferred_language) AS pref_lang FROM exaple_data AS a, UNNEST(preferred_language) AS pref_lang --그럼 unnest안에는 array만? struct는? SELECT student , MAX(IF(subject="수학", score, NULL)) AS 수학 , MAX(IF(subject="영어", score, NULL)) AS 영어 , MAX(IF(subject="과학", score, NULL)) AS 과학 FROM Table GROUP BY student ###팁 #같은 단어를 수정할 때,빨리하고 싶은 - 단어를 커서위에 올리고 커맨드 디 범위설정하고 수정하면 일괄수정 -> 인텔리데이에서는 어떻게 하지? #기대하는 아웃풋의 형태를 적어보는것 좋다 -> 쉐어포인트 컬럼에 만들기 프로젝트 시작전 - 어떤 업무를 함에 있어서 흐름을 아는 것이 중요하다(흐름을 모르면 어떤것을 왜 해야하는지 모를 수 있음) - 맥락 -> 목적 -> 퍼널 -> 가설 -> 분석 서비스의 목표 파악(어떤 문제를 해결하려고 하는지) 문제 정의: 핵심 문제 목표 정의 퍼널 정의 -> 우리도 이 데이터가 있는지 물어보기
1. ARRAY, STRUCT ### 1) array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요 # ARRAY : 같은 타입의 여러 데이터를 저장하고 싶을 때 # ARRAY를 Flatten(평면화) => UNNEST # UNNEST를 할 때는 CROSS JOIN + UNNEST(ARRAY_COLUMN) # UNNEST(ARRAY_COLUMN) AS 새로운 이름 # SELECT 절에서 새로운 이름으로 사용한다. 기존의 ARRAY_COLUMN은 사용하지 않는다! -- SELECT -- title -- , genre -- FROM `advanced.array_exercises` -- CROSS JOIN UNNEST(genres) AS genre -- ORDER BY 1, 2 ## 같은 결과를 출력하기 위해 정렬함. -- ; ### 2) array_exercises 테이블에서 각 영화(title)별로 배우(actor)와 배역(character)을 보여주세요. 배우와 배역은 별도의 컬럼으로 나와야 합니다. # 직접 접근하려면 actors = [ STRUCT(STRING, STRING)] # actors[SAFE_OFFSET(0)].actor # actors[SAFE_OFFSET(0)].character -- SELECT -- title -- , act.actor# AS actor -- , act.character# AS character -- FROM `advanced.array_exercises` -- CROSS JOIN UNNEST(actors) AS act -- ORDER BY 1 -- ; ### 3) array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르(genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다. # 데이터의 중복이 어느정도 생기는데, 그것은 어쩔 수 없는 이슈(CROSS JOIN) -- SELECT -- title -- -- actors, # ARRAY<STRUCT(STRING, STRING)> -- , act.actor# AS actor -- , act.character# AS character -- -- genres # ARRAY<STRING> -- , genre -- FROM `advanced.array_exercises` -- CROSS JOIN UNNEST(actors) AS act -- CROSS JOIN UNNEST(genres) AS genre -- -- WHERE 1=1 -- -- ## 강의 촬영 시점 이후에 수정된 듯 두 쿼리 모두 오류없이 실행 되는 것 같아요 ! -- -- AND act.actor = "Chris Evans" -- -- AND actor = "Chris Evans" -- ORDER BY 1 -- ; ### 4) 앱 로그 데이터(app_logs)의 배열을 풀어주세요. -- SELECT -- user_id -- , event_date -- , event_name -- , user_pseudo_id -- , evt_prm.key AS key -- , evt_prm.value.string_value AS string_value -- , evt_prm.value.int_value AS int_value -- FROM `advanced.app_logs` -- CROSS JOIN UNNEST(event_params) AS evt_prm -- WHERE 1=1 -- AND event_date = "2022-08-01" -- ORDER BY 2 -- ; ### WITH 문 변경 WITH base AS ( SELECT user_id , event_date , event_name , user_pseudo_id , evt_prm.key AS key , evt_prm.value.string_value AS string_value , evt_prm.value.int_value AS int_value FROM `advanced.app_logs` CROSS JOIN UNNEST(event_params) AS evt_prm WHERE 1=1 AND event_date = "2022-08-01" ) SELECT event_date , event_name , COUNT(DISTINCT user_id) AS cnt FROM base GROUP BY ALL ORDER BY cnt DESC 2. PIVOT # 1) orders 테이블에서 유저(user_id)별로 주문금액(amount)의 합계를 PIVOT해주세요. 날짜(order_date)를 행(Row)으로, user_id를 열(Column)으로 만들어야 합니다. -- 기대하는 output의 형태 -- order_date | user_1 | user_2 | user_3 -- PIVOT : MAX(IF(조건, TRUE일 때의 값, FALSE일 때의 값)) AS new_column + GROUP BY -- MAX 대신 집계 함수를 사용할 수도 있음. SUM -- FALSE일 때의 값은 NULL -- SELECT -- order_date -- , SUM(IF(user_id = 1, amount, 0)) AS user_1 -- , SUM(IF(user_id = 2, amount, 0)) AS user_2 -- , SUM(IF(user_id = 3, amount, 0)) AS user_3 -- FROM `advanced.orders` -- GROUP BY 1 -- ORDER BY 1 -- ; # 2) orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요.user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다. -- 기대하는 output의 형태 -- user_id | 2023-05-01 | 2023-05-02 | 2023-05-03 | 2023-05-04 | 2023-05-05 -- 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` -- 컬럼의 이름을 지정할 때, 영어 제외하고 backtick(`) -- ANY_VALUE : 그훕화 할 대상 중에 임의의 값을 선택한다 (NULL을 제외하고). ANY_VALUE에선 나머지 값들이 NULL이거나 확정적으로 값을 기대할 수 있을 때 사용한다! -- ANY_VALUE(IF(order_date="2023-05-01", amount, NULL)) AS `2023-05-01` -- FROM `advanced.orders` -- GROUP BY 1 -- ORDER BY 1 -- ; # 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. -- SELECT -- user_id -- , MAX(IF(order_date="2023-05-01", 1, 0)) AS `2023-05-01` -- , MAX(IF(order_date="2023-05-02", 1, 0)) AS `2023-05-02` -- , MAX(IF(order_date="2023-05-03", 1, 0)) AS `2023-05-03` -- , MAX(IF(order_date="2023-05-04", 1, 0)) AS `2023-05-04` -- , MAX(IF(order_date="2023-05-05", 1, 0)) AS `2023-05-05` -- FROM `advanced.orders` -- GROUP BY 1 -- ORDER BY 1 -- ; ## 앱 로그 PIVOT WITH base AS( SELECT -- * EXCEPT(event_params) # * EXCEPT(column) : 컬럼을 제외하고 다 보여줘! event_date , event_timestamp , event_name , user_id , user_pseudo_id , MAX(IF(param.key = "fierbase_screen", param.value.string_value, NULL)) AS fierbase_screen -- , MAX(IF(param.key = "food_id", param.value.string_value, NULL)) AS food_id # string_value엔 food_id 값들이 저장되어 있지 않음. , 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 sessioon_id FROM `advanced.app_logs` CROSS JOIN UNNEST(event_params) AS param WHERE 1=1 AND event_date = "2022-08-01" GROUP BY ALL ) SELECT event_date , COUNT(user_id) AS user_cnt FROM base WHERE 1=1 AND event_name = "click_cart" -- AND food_id = 1544 GROUP BY event_date 3. 퍼널 연습 문제 # 퍼널 분석 -- 퍼널 데이터 -- 우리가 사용할 이벤트 => 단계 -- - screen_view : welcome, home, food_category, restaurant, cart -- - click_payment -- step_number : 추후에 정렬을 위해 만들 것 -- 사용할 데이터 : 앱 로그 데이터, GA/Firebase => UNNEST => PIVOT -- 기간 : 2022-08-01 ~ 2022-08-18 WITH base as ( SELECT event_date , event_timestamp , event_name , user_id , user_pseudo_id , platform , 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 ), base2 as ( SELECT * , CONCAT(event_name, "-", firebase_screen) as event_screen FROM base WHERE 1=1 AND event_name IN ("screen_view", "click_payment") ), base3 as ( SELECT event_screen , event_date , CASE WHEN event_screen = "screen_view-welcome" THEN 1 WHEN event_screen = "screen_view-home" THEN 2 WHEN event_screen = "screen_view-food_category" THEN 3 WHEN event_screen = "screen_view-restaurant" THEN 4 WHEN event_screen = "screen_view-cart" THEN 5 WHEN event_screen = "click_payment-cart" THEN 6 ELSE NULL END as step_number , COUNT(DISTINCT user_pseudo_id) as cnt FROM base2 GROUP BY ALL HAVING step_number is not NULL ORDER BY event_date ) SELECT event_date , MAX(IF(base3.event_screen ="screen_view-welcome", cnt, NULL)) AS screen_view_welcome , MAX(IF(base3.event_screen ="screen_view-home", cnt, NULL)) AS screen_vie_home , MAX(IF(base3.event_screen ="screen_view-food_category", cnt, NULL)) AS screen_view_food_category , MAX(IF(base3.event_screen ="screen_view-restaurant", cnt, NULL)) AS screen_view_restaurant , MAX(IF(base3.event_screen ="screen_view-cart", cnt, NULL)) AS screen_view_cart FROM base3 GROUP BY ALL ORDER BY event_date
[ARRAY, STRUCT] 문제 1) array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요. SELECT title, genre FROM advanced.array_exercises CROSS JOIN UNNEST(genres) AS genre 쿼리 결과 1) 문제 2) array_exercises 테이블에서 각 영화(title)별로 배우(actor)와 배역(character)을 보여주세요. SELECT title, actor.actor, actor.character FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actor 쿼리 결과 2) 문제 3) array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르 (genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다. SELECT title, actor.actor AS actor, actor.character AS character, genre FROM advanced.array_exercises CROSS JOIN UNNEST(genres) AS genre CROSS JOIN UNNEST(actors) AS actor 쿼리 결과 3) 문제 4) 앱 로그 데이터(app_logs) 배열 풀기 SELECT event_date, event_timestamp, event_name, event_param.key AS key, event_param.value.string_value AS string_value, event_param.value.int_value AS int_value, user_id, user_pseudo_id, platform FROM `advanced.app_logs` CROSS JOIN UNNEST(event_params) AS event_param WHERE event_date = "2022-08-01" LIMIT 100 쿼리 결과 4) [PIVOT] 문제 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, SUM(IF(user_id = 1, amount, 0)) AS user_1, SUM(IF(user_id = 2, amount, 0)) AS user_2, SUM(IF(user_id = 3, amount, 0)) AS user_3 FROM advanced.orders GROUP BY ALL ORDER BY order_date 쿼리 결과 1) 문제 2) orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요. user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다 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 ALL ORDER BY user_id 쿼리 결과 2) 문제 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 SELECT user_id, MAX(IF(order_date = "2023-05-01", 1, 0)) AS `2023-05-01`, MAX(IF(order_date = "2023-05-02", 1, 0)) AS `2023-05-02`, MAX(IF(order_date = "2023-05-03", 1, 0)) AS `2023-05-03`, MAX(IF(order_date = "2023-05-04", 1, 0)) AS `2023-05-04`, MAX(IF(order_date = "2023-05-05", 1, 0)) AS `2023-05-05`, FROM advanced.orders GROUP BY ALL ORDER BY user_id 쿼리 결과 3) 문제 4)user_id = 32888이 카트 추가하기(click_cart)를 누를때 어떤 음식(food_id)을 담았나요? WITH base AS ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, 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 GROUP BY ALL ) SELECT user_id, event_date, COUNT(user_id) AS user_cnt, food_id FROM base WHERE user_id = 32888 and event_name = 'click_cart' GROUP BY ALL 쿼리 결과 4) [퍼널분석] WITH base AS ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, platform, MAX(IF(event_param.key = "firebase_screen", event_param.value.string_value, NULL)) AS firebase_screen, 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 ), 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") ) 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 쿼리 결과
1. ARRAY, STRUCT 1) SELECT title, movie_genres FROM advanced.array_exercises CROSS JOIN UNNEST(genres) AS movie_genres LIMIT 100 2) SELECT title, actor.actor, actor.character FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actor 3) SELECT title, actors.actor, actors.character, genres FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actors CROSS JOIN UNNEST(genres) AS genres WHERE actor = 'ChrisEvans' 4) select user_id , event_date , event_name , user_pseudo_id , param.key as key , param.value.string_value as string_value , param.value.int_value as int_value from advanced.app_logs , unnest(event_params) as param 2. PIVOT 1) select order_date , sum(if(user_id = 1,amount,0)) as user_1 , sum(if(user_id = 2,amount,0)) as user_2 , sum(if(user_id = 3,amount,0)) as user_3 from advanced.orders group by order_date order by order_date 2) select user_id , max(if(order_date = '2023-05-01',amount,0)) as `2023-05-01` , max(if(order_date = '2023-05-02',amount,0)) as `2023-05-02` , max(if(order_date = '2023-05-03',amount,0)) as `2023-05-03` , max(if(order_date = '2023-05-04',amount,0)) as `2023-05-04` , max(if(order_date = '2023-05-05',amount,0)) as `2023-05-05` from advanced.orders group by user_id order by user_id 3) select user_id , max(if(order_date = '2023-05-01',1,0)) as `2023-05-01` , max(if(order_date = '2023-05-02',1,0)) as `2023-05-02` , max(if(order_date = '2023-05-03',1,0)) as `2023-05-03` , max(if(order_date = '2023-05-04',1,0)) as `2023-05-04` , max(if(order_date = '2023-05-05',1,0)) as `2023-05-05` from advanced.orders group by user_id order by user_id 4) WITH base AS ( SELECT user_id, event_date, event_name, user_pseudo_id, event_param.key AS key, event_param.value.string_value AS string_value, event_param.value.int_value AS int_value FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event_param ) SELECT user_id, event_date, event_name, user_pseudo_id, MAX(IF(key = 'firebase_screen', string_value, NULL)) AS firebase_screen, MAX(IF(key = 'food_id', int_value, NULL)) AS food_id, MAX(IF(key = 'session_id', string_value, NULL)) AS session_id FROM base WHERE user_id = 32888 AND event_name = 'click_cart' GROUP BY ALL 3. 퍼널 1) with base as ( select event_date , event_name , event_timestamp , user_id , user_pseudo_id , platform , max(if(param.key = 'firebase_screen', param.value.string_value, null)) as firebase_screen from advanced.app_logs , unnest(event_params) as param where event_date between '2022-08-01' and '2022-08-18' group by all ), filter_event as ( select * except(event_name, firebase_screen) , concat(event_name, "-", firebase_screen) as event_name_with_screen from base where event_name in ('screen_view', 'click_payment') ) 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 group by all having step_number is not null order by event_date 2) with base as ( select event_date , event_name , event_timestamp , user_id , user_pseudo_id , platform , max(if(param.key = 'firebase_screen', param.value.string_value, null)) as firebase_screen from advanced.app_logs , unnest(event_params) as param where event_date between '2022-08-01' and '2022-08-18' group by all ), filter_event as ( select * except(event_name, firebase_screen) , concat(event_name, "-", firebase_screen) as event_name_with_screen from base where event_name in ('screen_view', 'click_payment') ), daily_group as ( 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 group by all having step_number is not null order by event_date ) select event_date , max(if(event_name_with_screen = 'screen_view-welcome',cnt,null)) as screen_view_welcome , max(if(event_name_with_screen = 'screen_view-home',cnt,null)) as screen_view_home , max(if(event_name_with_screen = 'screen_view-food_category',cnt,null)) as screen_view_food_category , max(if(event_name_with_screen = 'screen_view-restaurant',cnt,null)) as screen_view_restaurant , max(if(event_name_with_screen = 'screen_view-cart',cnt,null)) as screen_view_cart , max(if(event_name_with_screen = 'click_payment-cart',cnt,null)) as click_payment_cart from daily_group group by all order by event_date
ARRAY, STRUCT 연습 문제 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요. SELECT title, genre FROM advanced.array_exercises AS ae CROSS JOIN UNNEST(genres) AS genre 각 영화(title)별 배우(actor)와 배역(character)을 보여주세요.(별도 칼럼) SELECT title, actor.actor, actor.character FROM advanced.array_exercises AS ae CROSS JOIN UNNEST(actors) AS actor 각 영화(title)별로 배우(actor),배역(character),장르(genre)를 출력하세요. SELECT title, actor, character, genre FROM advanced.array_exercises AS ae CROSS JOIN UNNEST(actors) AS actor CROSS JOIN UNNEST(genres) AS genre 앱 로그 데이터(app_logs)의 배열을 풀어주세요. SELECT event_date, event_timestamp, event_name, user_pseudo_id, event_param.key AS key, event_param.value.string_value AS string_value, event_param.value.int_value AS int_value FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event_param PIVOT 연습 문제 유저(user_id)별로 주문 금액(amount)의 합계를 PIVOT (날짜를 행, user_id를 열) SELECT order_date, SUM(IF(user_id = 1, amount, 0)) AS user_1, SUM(IF(user_id = 2, amount, 0)) AS user_2, SUM(IF(user_id = 3, amount, 0)) AS user_3 FROM advanced.orders GROUP BY order_date ORDER BY order_date 날짜별로 유저들의 주문금액의 합계를 PIVOT (user_id를 행, order_date를 열) 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 사용자별, 날짜별로 주문이 있다면 1, 없다면 0으로 PIVOT 해주세요 (user_id를 행, order_date를 열) SELECT user_id, IF(SUM(IF(order_date = '2023-05-01', amount, 0))>0,1,0) AS `2023-05-01`, IF(SUM(IF(order_date = '2023-05-02', amount, 0))>0,1,0) AS `2023-05-03`, IF(SUM(IF(order_date = '2023-05-03', amount, 0))>0,1,0) AS `2023-05-02`, IF(SUM(IF(order_date = '2023-05-04', amount, 0))>0,1,0) AS `2023-05-04`, IF(SUM(IF(order_date = '2023-05-05', amount, 0))>0,1,0) AS `2023-05-05` FROM advanced.orders GROUP BY user_id ORDER BY user_id user_id = 32888이 카트 추가하기(click_cart)를 누를때 어떤 음식(food_id)을 담았나요? SELECT user_id, 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 FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event_param WHERE user_id = 32888 AND event_name = 'click_cart' GROUP BY user_id, event_timestamp -- 카트에 담은 음식(food_id): 1559, 1942 퍼널 쿼리 연습 문제 일자별 이벤트 집계 후 PIVOT WITH funnel_data AS ( SELECT *, 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 END AS step_number FROM ( SELECT event_date, event_timestamp, user_pseudo_id, concat(event_name, '-', event_param.value.string_value) AS event_name_with_screen FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event_param WHERE event_param.key = 'firebase_screen' ) AS unnested_app_logs WHERE event_name_with_screen IN ( 'screen_view-welcome', 'screen_view-home', 'screen_view-food_category', 'screen_view-restaurant', 'screen_view-cart', 'click_payment-cart' ) )SELECT event_date, COUNT(IF(step_number = 1, user_pseudo_id, NULL)) AS `screen_view-welcome`, COUNT(IF(step_number = 2, user_pseudo_id, NULL)) AS `screen_view-home`, COUNT(IF(step_number = 3, user_pseudo_id, NULL)) AS `screen_view-food_category`, COUNT(IF(step_number = 4, user_pseudo_id, NULL)) AS `screen_view-restaurant`, COUNT(IF(step_number = 5, user_pseudo_id, NULL)) AS `screen_view-cart`, COUNT(IF(step_number = 6, user_pseudo_id, NULL)) AS `click_payment-cart` FROM funnel_data GROUP BY event_date ORDER BY event_date
1. ARRAY, STRUCT 연습문제 문제 1) array_exercise테이블에서 각 영화(title)별로 장르(genres)를 UNNEST 해서 보여주세요 SELECT title, genres FROM `analystic-project.advanced.array_exercises` , UNNEST(genres) AS genres ; 문제 2) array_exercises 테이블에서 각 영화(title)별로 배우(actor)와 배역(character)을 보여주세요. 배우와 배역은 별도의 컬럼으로 나와야합니다 SELECT title, actors.actor, actors.character FROM `analystic-project.advanced.array_exercises` , UNNEST(actors) AS actors ; 문제 3) array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르 (genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다 SELECT title, actors.actor, actors.character, genres FROM `analystic-project.advanced.array_exercises` , UNNEST(actors) AS actors, UNNEST(genres) genres ; 문제 4) 앱 로그 데이터(app_logs) 배열 풀기 SELECT user_id, event_date, event_name, user_pseudo_id, pr.key, pr.value.string_value, pr.value.int_value FROM `analystic-project.advanced.app_logs` , UNNEST(event_params) AS pr WHERE event_date = "2022-08-01" LIMIT 1000 ; 2. PIVOT 연습문제 풀이 문제 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, COALESCE(SUM(IF(user_id = 1, amount, null)),0) AS user_1, COALESCE(SUM(IF(user_id = 2, amount, null)),0) AS user_2, COALESCE(SUM(IF(user_id = 3, amount, null)),0) AS user_3 FROM advanced.orders GROUP BY order_date ORDER BY order_date ; 문제 2) orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요. user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다 SELECT user_id, COALESCE(SUM(IF(order_date = '2023-05-01', amount, null)),0) AS `2023-05-01`, COALESCE(SUM(IF(order_date = '2023-05-02', amount, null)),0) AS `2023-05-02`, COALESCE(SUM(IF(order_date = '2023-05-03', amount, null)),0) AS `2023-05-03`, COALESCE(SUM(IF(order_date = '2023-05-04', amount, null)),0) AS `2023-05-04`, COALESCE(SUM(IF(order_date = '2023-05-05', amount, null)),0) AS `2023-05-05`, FROM advanced.orders GROUP BY user_id ORDER BY user_id ; 문제 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 SELECT user_id, MAX(IF(order_date = '2023-05-01' AND order_id is not null, 1, 0)) AS `2023-05-01`, MAX(IF(order_date = '2023-05-02' AND order_id is not null, 1, 0)) AS `2023-05-02`, MAX(IF(order_date = '2023-05-03' AND order_id is not null, 1, 0)) AS `2023-05-03`, MAX(IF(order_date = '2023-05-04' AND order_id is not null, 1, 0)) AS `2023-05-04`, MAX(IF(order_date = '2023-05-05' AND order_id is not null, 1, 0)) AS `2023-05-05`, FROM advanced.orders GROUP BY user_id ORDER BY user_id ; 문제 4) user_id = 32888이 카트 추가하기(click_cart)를 누를때 어떤 음식(food_id)을 담았나요? WITH app_order_raw AS ( SELECT user_id, event_date, event_name, user_pseudo_id, pr.key, pr.value.string_value, pr.value.int_value FROM advanced.app_logs, UNNEST(event_params) AS pr WHERE event_date = '2022-08-01' ) SELECT user_id, event_date, event_name, user_pseudo_id, MAX(IF(key = 'firebase_screen', string_value, null)) AS firebase_screen, MAX(IF(key = 'food_id', int_value, null)) AS food_id, MAX(IF(key = 'session_id', string_value, null)) AS session_id, FROM app_order_raw GROUP BY user_id, event_date, event_name, user_pseudo_id ; 3. 퍼널분석 문제 1) 각 퍼널의 유저 수를 집계 / 데이터 기준: 2022-08-01 ~ 2022-08-18 WITH funnel_data_raw AS ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, MAX(IF(pr.key = 'firebase_screen', pr.value.string_value, null)) AS screen_name, CONCAT(event_name, '-', MAX(IF(pr.key = 'firebase_screen', pr.value.string_value, null))) AS event_name_with_screen FROM advanced.app_logs, UNNEST(event_params) AS pr WHERE event_date BETWEEN '2022-08-01' AND '2022-08-18' GROUP BY 1,2,3,4,5 ) SELECT 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 END AS step_number, COUNT(DISTINCT user_pseudo_id) AS cnt FROM funnel_data_raw WHERE event_name IN ('screen_view', 'click_payment') AND screen_name IN ('welcome', 'home', 'food_category', 'restaurant', 'cart') GROUP BY 1,2 ORDER BY 2 ; 문제 2) 일자별 퍼널 유저 수 집계 WITH funnel_data_raw AS ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, MAX(IF(pr.key = 'firebase_screen', pr.value.string_value, null)) AS screen_name, CONCAT(event_name, '-', MAX(IF(pr.key = 'firebase_screen', pr.value.string_value, null))) AS event_name_with_screen FROM advanced.app_logs, UNNEST(event_params) AS pr WHERE event_date BETWEEN '2022-08-01' AND '2022-08-18' GROUP BY 1,2,3,4,5 ) 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 END AS step_number, COUNT(DISTINCT user_pseudo_id) AS cnt FROM funnel_data_raw WHERE event_name IN ('screen_view', 'click_payment') AND screen_name IN ('welcome', 'home', 'food_category', 'restaurant', 'cart') GROUP BY 1,2,3 ORDER BY 1,3 ; 문제 3) 일자별 퍼널 유저 수 집계 형태를 PIVOT형태로 전환하기 WITH funnel_data_raw AS ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, MAX(IF(pr.key = 'firebase_screen', pr.value.string_value, null)) AS screen_name, CONCAT(event_name, '-', MAX(IF(pr.key = 'firebase_screen', pr.value.string_value, null))) AS event_name_with_screen FROM advanced.app_logs, UNNEST(event_params) AS pr WHERE event_date BETWEEN '2022-08-01' AND '2022-08-18' GROUP BY 1,2,3,4,5 ), daily_funnel_user_count as ( 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 END AS step_number, COUNT(DISTINCT user_pseudo_id) AS cnt FROM funnel_data_raw WHERE event_name IN ('screen_view', 'click_payment') AND screen_name IN ('welcome', 'home', 'food_category', 'restaurant', 'cart') GROUP BY 1,2,3 ORDER BY 1,3 ) SELECT event_date, MAX(IF(step_number = 1, cnt, null)) AS `screen_view-welcome`, MAX(IF(step_number = 2, cnt, null)) AS `screen_view-home`, MAX(IF(step_number = 3, cnt, null)) AS `screen_view-food_category`, MAX(IF(step_number = 4, cnt, null)) AS `screen_view-restaurant`, MAX(IF(step_number = 5, cnt, null)) AS `screen_view-cart`, MAX(IF(step_number = 6, cnt, null)) AS `click_payment-cart`, FROM daily_funnel_user_count GROUP BY ALL ORDER BY 1 ;
<PART 1> ARRAY, STRUCT 연습문제 Q1. array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요. -- 출제의도: 배열 UNNEST의 기본 형태를 사용할 수 있는가? SELECT title , genre FROM advanced.array_exercises CROSS JOIN UNNEST(genres) AS genre ORDER BY title; Q2. array_exercise 테이블에서 각 영화(title)별로 배우(actors)와 배역(character)을 보여주세요. 배우와 배역은 별도의 컬럼으로 나와야 합니다. -- 출제의도: 다중 배열 구조에서 UNNEST를 사용할 수 있는가? SELECT title , actor.actor , actor.character FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actor ORDER BY title; Q3. array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르(genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다. -- 출제의도: 여러 칼럼을 동시에 UNNEST할 수 있는가? SELECT title , actor.actor , actor.character , genre FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actor CROSS JOIN UNNEST(genres) AS genre ORDER BY title; Q4. 앱 로그 데이터(app_logs)의 배열을 풀어주세요. -- 출제의도: 다중 struct 구조의 데이터를 평면화하여 쿼리로 호출할 수 있는가? SELECT user_id , event_date , event_name , user_pseudo_id , event.key , event.value.string_value , event.value.int_value FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event WHERE event_date = '2022-08-01'; <PART 2> PIVOT 연습문제 Q1. orders 테이블에서 유저(user_id)별로 주문 금액(amount)의 합계를 PIVOT 해주세요. 날짜(order_date)를 행(Row)으로, user_id를 열(Column)으로 만들어야 합니다. -- 출제의도: 집계 함수와 조건 함수를 결합하여 PIVOT 테이블을 만들 수 있는가? SELECT order_date , SUM(IF(user_id=1, amount, 0)) AS user_1 , SUM(IF(user_id=2, amount, 0)) AS user_2 , SUM(IF(user_id=3, amount, 0)) AS user_3 FROM advanced.orders GROUP BY ALL ORDER BY order_date; Q2. orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요. user_id 를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다. -- 출제의도 : PIVOT 테이블 구성 시, 날짜 칼럼을 이용하여 시계열 방식을 구성할 수 있는가? 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 ALL ORDER BY user_id; Q3. 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. -- 출제의도 : PIVOT 테이블 구성 시, 집계 함수로 MAX를 사용할 수 있는가? SELECT user_id , MAX(IF(order_date = '2023-05-01', 1, 0)) AS `2023-05-01` , MAX(IF(order_date = '2023-05-02', 1, 0)) AS `2023-05-02` , MAX(IF(order_date = '2023-05-03', 1, 0)) AS `2023-05-03` , MAX(IF(order_date = '2023-05-04', 1, 0)) AS `2023-05-04` , MAX(IF(order_date = '2023-05-05', 1, 0)) AS `2023-05-05` FROM advanced.orders GROUP BY ALL ORDER BY user_id; Q4. user_id = 32888 이 카트 추가하기(click_cart)를 누를 때 어떤 음식 (food_id)을 담았는지 구해주세요. key 를 Column 으로 두고, string_value 나 int_value를 Column의 값으로 설정해서 풀어주세요. -- 출제의도 : PIVOT 테이블을 앱로그 데이터에 사용하여, 조건문으로 개별 유저 데이터를 특정할 수 있는가? WITH base AS( SELECT event_date , event_timestamp , event_name , user_id , user_pseudo_id , MAX(IF(params.key = 'firebase_screen', params.value.string_value, NULL)) AS firebase_screen , MAX(IF(params.key = 'food_id', params.value.int_value, NULL)) AS food_id , MAX(IF(params.key = 'session_id', params.value.string_value, NULL)) AS session_id FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS params WHERE event_date = '2022-08-01' GROUP BY ALL ) SELECT * FROM base WHERE event_name = 'click_cart' and user_id = 32888 -- 실행결과 : food_id = 1942 <PART 3> 퍼널 연습문제 -- 출제의도: 앱 로그 데이터에서 원하는 이벤트를 추출해, 퍼널 분석을 위한 전처리를 진행할 수 있는가? -- step 1. UNNEST를 통한 base 데이터 준비 WITH base AS( SELECT event_date , event_timestamp , event_name , event.key AS event_key , event.value.string_value AS event_string_value , event.value.int_value AS event_int_value , user_id , user_pseudo_id , platform FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event WHERE event_date BETWEEN '2022-08-01' AND '2022-08-22' ), -- step 2. 필요한 퍼널 이벤트에만 step_number를 세팅하여 준비 sorted_events AS( SELECT event_date , CONCAT(event_name, "-", event_string_value) AS event_name_with_screen , CASE WHEN event_name = 'screen_view' AND event_string_value = 'welcome' THEN 1 WHEN event_name = 'screen_view' AND event_string_value = 'home' THEN 2 WHEN event_name = 'screen_view' AND event_string_value = 'food_category' THEN 3 WHEN event_name = 'screen_view' AND event_string_value = 'restaurant' THEN 4 WHEN event_name = 'screen_view' AND event_string_value = 'cart' THEN 5 WHEN event_name = 'click_payment' AND event_string_value = 'cart' THEN 6 ELSE NULL END AS step_number , user_pseudo_id FROM base WHERE event_key = 'firebase_screen' ) -- step 3. 최종 조회 쿼리 SELECT event_date , event_name_with_screen , step_number , COUNT(DISTINCT user_pseudo_id) AS cnt FROM sorted_events WHERE step_number IS NOT NULL GROUP BY ALL ORDER BY event_date, step_number
ARRAY -- 1) array_exercises 테이블에서 각 영화(title)별로 장르(genres)를 UNNEST해서 보여주세요. SELECT title, genre FROM advanced.array_exercises CROSS JOIN UNNEST(genres) AS genre --2) array_exercises 테이블에서 각 영화(title)별로 배우(actor)와 배역(character)을 보여주세요. 배우와 배역은 별도의 컬럼으로 나와야 합니다. SELECT title, actor.actor, actor.character FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actor --3) array_exercises 테이블에서 각 영화(title)별로 배우(actor), 배역(character), 장르(genre)를 출력하세요. 한 Row에 배우, 배역, 장르가 모두 표시되어야 합니다. SELECT title, actor.actor AS actor, actor.character AS character, genre FROM advanced.array_exercises CROSS JOIN UNNEST(actors) AS actor CROSS JOIN UNNEST(genres) AS genre WHERE actor.actor = 'Chris Evans' AND genre = 'Action'es) AS genre -- 4) 앱 로그 데이터(app_logs)의 배열을 풀어주세요 WITH base AS ( SELECT user_id, event_date, event_name, user_pseudo_id, event_param.key AS key, event_param.value.string_value AS string_value, event_param.value.int_value AS int_value FROM advanced.app_logs CROSS JOIN UNNEST(event_params) AS event_param WHERE event_date = '2022-08-01') PIVOT --1. orders 테이블에서 유저(user_id)별로 주문 금액(amount)의 합계를 PIVOT 해주세요. 날짜(order_date)를 행(Row)으로, user_id를 열(Column)으로 만들어아 합니다. SELECT order_date, MAX(IF(user_id = 1, sum_of_amount, 0)) AS user_1, MAX(IF(user_id = 2, sum_of_amount, 0)) AS user_2, MAX(IF(user_id = 3, sum_of_amount, 0)) AS user_3 FROM ( SELECT order_date, user_id, SUM(amount) AS sum_of_amount FROM advanced.orders GROUP BY order_date, user_id ) GROUP BY order_date ORDER BY order_date --2.orders 테이블에서 날짜(order_date)별로 유저들의 주문 금액(amount)의 합계를 PIVOT 해주세요. user_id를 행(Row)으로, order_date를 열(Column)으로 만들어야 합니다. SELECT user_id, MAX(IF(order_date = "2023-05-01", amount, 0)) AS `2023-05-01`, MAX(IF(order_date = "2023-05-02", amount, 0)) AS `2023-05-02`, MAX(IF(order_date = "2023-05-03", amount, 0)) AS `2023-05-03`, MAX(IF(order_date = "2023-05-04", amount, 0)) AS `2023-05-04`, MAX(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. SELECT user_id, MAX(IF(order_date = '2023-05-01', 1, 0)) AS `2023-05-01`, MAX(IF(order_date = '2023-05-02', 1, 0)) AS `2023-05-02`, MAX(IF(order_date = '2023-05-03', 1, 0)) AS `2023-05-03`, MAX(IF(order_date = '2023-05-04', 1, 0)) AS `2023-05-04`, MAX(IF(order_date = '2023-05-05', 1, 0)) AS `2023-05-05`, FROM advanced.orders GROUP BY user_id 퍼널 WITH base AS ( SELECT event_date, event_timestamp, event_name, user_id, user_pseudo_id, platform, MAX(IF(event_param.key = "firebase_screen", event_param.value.string_value, NULL)) AS firebase_screen, 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 ), 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") ) 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 아직 SQL 익숙지 않아서, 강의 들으면서 코드를 이해하려고 했습니다. 얼른 빅쿼리 SQL입문 강의도 다 듣고, 2주차에 더 실력이 올라갔으면 좋겠습니다!
