Files
homework/itikhomirov/clickhouse/Pet_project.sql
T
2025-12-08 22:52:38 +03:00

283 lines
7.4 KiB
SQL

CREATE database pet;
-- Этап 1
-- Задание 1
-- Перенести данные как в пет-проекте с Greenplum
-- Создаем таблицы, чтобы потом загрузить в них данные из таблиц GreenPlum
-- Таблица user из GP
CREATE TABLE pet.core_user (
user_id Int32,
name String,
age Int32,
email String,
registration_date DateTime64(3, 'UTC')
)
ENGINE = MergeTree
ORDER BY user_id;
-- Таблица course из GP
CREATE TABLE pet.core_course (
course_id Int32,
title String,
category String
)
ENGINE = MergeTree
ORDER BY course_id;
-- Таблица lesson из GP
CREATE TABLE pet.core_lesson (
lesson_id Int32,
title String,
duration_min Int32,
course_id Int32
)
ENGINE = MergeTree
ORDER BY lesson_id;
-- Таблица fact_lesson_views из GP
CREATE TABLE pet.core_fact_lesson_views (
user_id Int32,
lesson_id Int32,
course_id Int32,
viewed_at DateTime64(3, 'UTC')
)
ENGINE = MergeTree
ORDER BY user_id;
-- Таблица enrollments из GP
CREATE TABLE pet.core_enrollments (
user_id Int32,
course_id Int32,
enrolled_at DateTime64(3, 'UTC')
)
ENGINE = MergeTree
ORDER BY user_id;
-- Этап 2. Создание витринных таблиц
-- Задание 2: Подготовь структуру витрины lesson_popularity_summary
-- Создаем пустую таблицу lesson_popularity_summary(структуру витрины)
CREATE TABLE pet.lesson_popularity_summary
(
lesson_id UInt32,
lesson_title String,
course_id UInt32,
course_title String,
total_views UInt32,
unique_users UInt32,
first_view DateTime64(3, 'UTC'),
last_view DateTime64(3, 'UTC')
)
ENGINE = MergeTree
ORDER BY (lesson_id, course_id);
-- Создаем Materialized View
CREATE MATERIALIZED VIEW pet.mv_lesson_popularity_summary
TO pet.lesson_popularity_summary
AS
SELECT
flv.lesson_id AS lesson_id,
l.title AS lesson_title,
flv.course_id AS course_id,
c.title AS course_title,
count() AS total_views,
uniqExact(flv.user_id) AS unique_users,
min(flv.viewed_at) AS first_view,
max(flv.viewed_at) AS last_view
FROM pet.core_fact_lesson_views AS flv
JOIN pet.core_lesson AS l ON flv.lesson_id = l.lesson_id
JOIN pet.core_course AS c ON flv.course_id = c.course_id
GROUP BY
flv.lesson_id,
l.title,
flv.course_id,
c.title;
-- Вставляем данные в таблицу pet.lesson_popularity_summary
INSERT INTO pet.lesson_popularity_summary
SELECT
flv.lesson_id,
l.title AS lesson_title,
flv.course_id,
c.title AS course_title,
count() AS total_views,
uniqExact(flv.user_id) AS unique_users,
min(flv.viewed_at) AS first_view,
max(flv.viewed_at) AS last_view
FROM pet.core_fact_lesson_views AS flv
JOIN pet.core_lesson AS l ON flv.lesson_id = l.lesson_id
JOIN pet.core_course AS c ON flv.course_id = c.course_id
GROUP BY
flv.lesson_id,
l.title,
flv.course_id,
c.title;
-- Задание 3: Создай витрину inactive_users_summary
-- Создаем пустую таблицу inactive_users_summary(структуру витрины)
CREATE TABLE pet.inactive_users_summary
(
user_id UInt32,
name String,
email String,
age UInt8,
registered_courses_count UInt32,
registration_date DateTime64(3, 'UTC')
)
ENGINE = MergeTree
ORDER BY (user_id);
-- Создаем Materialized View
CREATE MATERIALIZED VIEW pet.mv_inactive_users_summary
TO pet.inactive_users_summary
AS
SELECT
u.user_id AS user_id,
u.name AS name,
u.email AS email,
u.age AS age,
countDistinct(e.course_id) AS registered_courses_count,
u.registration_date AS registration_date
FROM pet.core_user AS u
LEFT JOIN pet.core_enrollments AS e ON u.user_id = e.user_id
WHERE u.user_id NOT IN (
SELECT DISTINCT user_id
FROM pet.core_fact_lesson_views
)
GROUP BY
u.user_id,
u.name,
u.email,
u.age,
u.registration_date;
-- Вставляем данные в таблицу inactive_users_summary
INSERT INTO pet.inactive_users_summary
SELECT
u.user_id,
u.name,
u.email,
u.age,
countDistinct(e.course_id) AS registered_courses_count,
u.registration_date
FROM pet.core_user AS u
LEFT JOIN pet.core_enrollments AS e
ON u.user_id = e.user_id
WHERE u.user_id NOT IN (
SELECT DISTINCT user_id
FROM pet.core_fact_lesson_views
)
/*
* этот подзапрос WHERE... просто выбрасывает всех, кто хоть раз что‑то посмотрел,
* независимо от количества строк и особенностей JOIN’ов.
*/
GROUP BY
u.user_id,
u.name,
u.email,
u.age,
u.registration_date;
-- Задание 4: Создай витрину course_completion_rate
CREATE TABLE pet.course_completion_rate
(
user_id UInt32,
user_name String,
course_id UInt32,
course_title String,
lessons_in_course UInt32,
lessons_viewed UInt32,
completion_rate Float32
)
ENGINE = MergeTree
ORDER BY (user_id, course_id);
-- Создаем Materialized View
CREATE MATERIALIZED VIEW pet.mv_course_completion_rate
TO pet.course_completion_rate
AS
SELECT
u.user_id AS user_id,
u.name AS user_name,
c.course_id AS course_id,
c.title AS course_title,
countDistinct(l.lesson_id) AS lessons_in_course,
countDistinct(flv.lesson_id) AS lessons_viewed,
if(countDistinct(l.lesson_id) = 0,0,
round(countDistinct(flv.lesson_id) * 1.0 / countDistinct(l.lesson_id),2)
) AS completion_rate
FROM pet.core_enrollments AS e
JOIN pet.core_user AS u ON e.user_id = u.user_id
JOIN pet.core_course AS c ON e.course_id = c.course_id
JOIN pet.core_lesson AS l ON c.course_id = l.course_id
LEFT JOIN pet.core_fact_lesson_views AS flv
ON flv.user_id = e.user_id
AND flv.course_id = e.course_id
GROUP BY
u.user_id,
u.name,
c.course_id,
c.title;
-- Вставляем данные в таблицу pet.course_completion_rate
INSERT INTO pet.course_completion_rate
-- Создаем две временные таблицы:
-- 1) Сколько уроков в каждом курсе
WITH
lessons_per_course AS (
SELECT
course_id,
countDistinct(lesson_id) AS lessons_in_course
FROM pet.core_lesson
GROUP BY course_id
),
-- 2) Сколько уроков каждого курса посмотрел пользователь
views_per_user_course AS (
SELECT
user_id,
course_id,
countDistinct(lesson_id) AS lessons_viewed
FROM pet.core_fact_lesson_views
GROUP BY user_id, course_id
)
-- Основной запрос
SELECT
u.user_id,
u.name AS user_name,
c.course_id,
c.title AS course_title,
lpc.lessons_in_course,
coalesce(vpc.lessons_viewed, 0) AS lessons_viewed,
ROUND(
COALESCE(
NULLIF(coalesce(vpc.lessons_viewed, 0) * 1.0, 0)
/ NULLIF(lpc.lessons_in_course * 1.0, 0),
0
),
2
) AS completion_rate
FROM pet.core_user AS u
JOIN pet.core_enrollments AS e ON u.user_id = e.user_id
JOIN pet.core_course AS c ON e.course_id = c.course_id
JOIN lessons_per_course AS lpc ON c.course_id = lpc.course_id
LEFT JOIN views_per_user_course AS vpc ON vpc.user_id = u.user_id AND vpc.course_id = c.course_id
ORDER BY
u.user_id,
c.course_id;