Пет-проект: Витринный слой. Задание 4: Создай витрину course_completion_rate

2026-05-20 15:22:48 +03:00
parent 1da82b6877
commit 18cfc2a077
1 changed files with 117 additions and 0 deletions
@@ -0,0 +1,117 @@
+--DROP TABLE mnoskov.course_completion_rate ON CLUSTER clickhouse_cluster sync;
+CREATE TABLE mnoskov.course_completion_rate
+ON CLUSTER clickhouse_cluster
+(
+	user_id UInt64,
+	user_name String,
+	course_id UInt64,
+	course_title String,
+	lessons_in_course UInt16,
+	lesson_viewed UInt16,
+	completion_rate Float32
+)
+ENGINE = ReplicatedMergeTree(
+    '/clickhouse/tables/{shard}/mnoskov/course_completion_rate',
+    '{replica}'
+)
+ORDER BY (user_id, course_id);
+
+--DROP VIEW mnoskov.mv_course_completion_rate ON CLUSTER clickhouse_cluster sync;
+CREATE MATERIALIZED VIEW mnoskov.mv_course_completion_rate
+ON CLUSTER clickhouse_cluster
+TO mnoskov.course_completion_rate
+AS
+SELECT
+	u.user_id AS user_id,
+	u.name AS user_name,
+	cl.course_id AS course_id,
+	cl.course_title AS course_title,
+	cl.lessons_in_course AS lessons_in_course,
+	lv.lesson_viewed AS lesson_viewed,
+	ifnull(round(toFloat32(ifnull(lv.lesson_viewed, 0)) / nullif(cl.lessons_in_course, 0), 2), 0) AS completion_rate -- вроде все логично
+FROM
+	mnoskov.core_enrollments e
+RIGHT JOIN
+	mnoskov.core_dim_user u ON e.user_id = u.user_id
+JOIN (
+	SELECT
+		c.course_id,
+		c.title AS course_title,
+		uniq(lesson_id) lessons_in_course
+	FROM
+		mnoskov.core_dim_course c
+	JOIN
+		mnoskov.core_dim_lesson l ON l.course_id = c.course_id
+	GROUP BY
+		c.course_id, c.title
+) cl ON e.course_id = cl.course_id
+LEFT JOIN (
+	SELECT
+		user_id,
+		course_id,
+		uniq(lesson_id) AS lesson_viewed
+	FROM
+		mnoskov.core_fact_lesson_views
+	GROUP BY
+		user_id, course_id
+) lv ON e.user_id = lv.user_id AND e.course_id = lv.course_id
+ORDER BY
+	u.user_id, e.course_id;
+
+-- Чтение очищение заполнение
+SHOW TABLES FROM mnoskov;
+
+SELECT * FROM mnoskov.core_enrollments;
+SELECT * FROM mnoskov.course_completion_rate;
+
+TRUNCATE TABLE mnoskov.core_enrollments ON CLUSTER clickhouse_cluster sync;
+TRUNCATE TABLE mnoskov.course_completion_rate ON CLUSTER clickhouse_cluster sync;
+
+INSERT INTO mnoskov.core_enrollments
+(user_id, course_id, enrolled_at)
+VALUES
+(6, 1, '2023-01-05 23:00:00'),
+(14, 2, '2023-01-08 23:00:00'),
+(14, 3, '2023-01-08 23:00:00'),
+(14, 1, '2023-01-08 23:00:00'),
+(6, 2, '2023-01-09 23:00:00'),
+(15, 1, '2023-01-10 23:00:00'),
+(7, 1, '2023-01-12 23:00:00'),
+(1, 1, '2023-01-14 23:00:00'),
+(12, 1, '2023-01-15 23:00:00'),
+(13, 2, '2023-01-18 23:00:00'),
+(2, 1, '2023-01-19 23:00:00'),
+(8, 3, '2023-01-21 23:00:00'),
+(8, 1, '2023-01-21 23:00:00'),
+(8, 2, '2023-01-21 23:00:00'),
+(1, 2, '2023-01-31 23:00:00'),
+(7, 2, '2023-01-31 23:00:00'),
+(14, 4, '2023-01-31 23:00:00'),
+(15, 2, '2023-01-31 23:00:00'),
+(6, 4, '2023-02-04 23:00:00'),
+(1, 4, '2023-02-09 23:00:00'),
+(9, 4, '2023-02-11 23:00:00'),
+(9, 1, '2023-02-14 23:00:00'),
+(16, 2, '2023-02-15 23:00:00'),
+(16, 4, '2023-02-15 23:00:00'),
+(13, 4, '2023-02-19 23:00:00'),
+(3, 1, '2023-02-24 23:00:00'),
+(3, 2, '2023-03-04 23:00:00'),
+(15, 3, '2023-03-04 23:00:00'),
+(6, 3, '2023-03-07 23:00:00'),
+(4, 2, '2023-03-09 23:00:00'),
+(7, 3, '2023-03-09 23:00:00'),
+(12, 3, '2023-03-09 23:00:00'),
+(2, 3, '2023-03-14 23:00:00'),
+(5, 5, '2023-03-19 23:00:00'),
+(14, 5, '2023-03-19 23:00:00'),
+(17, 3, '2023-03-21 23:00:00'),
+(17, 5, '2023-03-24 23:00:00'),
+(4, 3, '2023-03-31 23:00:00'),
+(18, 1, '2023-04-01 23:00:00'),
+(18, 4, '2023-04-04 23:00:00'),
+(19, 1, '2023-04-10 23:00:00'),
+(19, 2, '2023-04-10 23:00:00'),
+(10, 1, '2023-04-15 23:00:00'),
+(20, 2, '2023-04-18 23:00:00'),
+(11, 2, '2023-04-20 23:00:00');