Merge pull request 'OS-330' (#45) from OS-330 into main

Reviewed-on: https://gitea.de-infra.servehttp.com/de-infrastructure/DWH/pulls/45
2026-01-14 15:13:44 +04:00
parent 374a86ba2c ebe1f00716
commit 08e09f7515
2 changed files with 283 additions and 0 deletions
@@ -0,0 +1,215 @@
 CREATE TABLE IF NOT EXISTS amaksatov.users (
    id SERIAL PRIMARY KEY,
    name VARCHAR(100),
    age INT,
    email VARCHAR(255),
    registration_date DATE
 );
 CREATE TABLE IF NOT EXISTS amaksatov.courses (
    id SERIAL PRIMARY KEY,
    title VARCHAR(255),
    category VARCHAR(100),
    created_at DATE
 );
 CREATE TABLE IF NOT EXISTS amaksatov.lessons (
    id SERIAL PRIMARY KEY,
    course_id INT,
    title VARCHAR(255),
    duration_min INT,
    FOREIGN KEY (course_id) REFERENCES amaksatov.courses(id)
 );
 CREATE TABLE IF NOT EXISTS amaksatov.enrollments (
    id SERIAL PRIMARY KEY,
    user_id INT,
    course_id INT,
    enrolled_at date,
    FOREIGN KEY (user_id) REFERENCES amaksatov.users(id),
    FOREIGN KEY (course_id) REFERENCES amaksatov.courses(id)
 );
 CREATE TABLE IF NOT EXISTS amaksatov.lesson_views (
    id SERIAL PRIMARY KEY,
    user_id INT,
    lesson_id INT,
    viewed_at TIMESTAMP,
    FOREIGN KEY (user_id) REFERENCES amaksatov.users(id),
    FOREIGN KEY (lesson_id) REFERENCES amaksatov.lessons(id)
 );
 INSERT
    INTO
    amaksatov.users (
        name,
        age,
        email,
        registration_date
    )
 VALUES (
    'Alice',
    25,
    'alice@mail.com',
    '2023-01-10'
 ),
 (
    'Bob',
    30,
    'bob@gmail.com',
    '2023-02-05'
 ),
 (
    'Charlie',
    22,
    'charlie@mail.com',
    '2023-02-20'
 ),
 (
    'Diana',
    28,
    'diana@mail.com',
    '2023-03-01'
 ),
 (
    'Ethan',
    35,
    'ethan@gmail.com',
    '2023-03-10'
 );
 SELECT * FROM amaksatov.users;
 INSERT
    INTO
    amaksatov.courses (
        title,
        category,
        created_at
    )
 VALUES 
 (
    'SQL для начинающих',
    'data',
    '2023-01-01'
 ),
 (
    'Python для анализа данных',
    'programming',
    '2023-01-15'
 ),
 (
    'BI с нуля',
    'business',
    '2023-03-01'
 );
 INSERT
    INTO
    amaksatov.lessons (
        course_id,
        title,
        duration_min
    )
 VALUES
 (
    1,
    'SELECT и FROM',
    10
 ),
 (
    1,
    'JOIN',
    15
 ),
 (
    2,
    'Pandas',
    20
 ),
 (
    3,
    'Основы BI',
    12
 ),
 (
    3,
    'Метрики и дашборды',
    18
 );
 INSERT
    INTO
    amaksatov.enrollments (
        user_id,
        course_id,
        enrolled_at
    )
 VALUES
 (
    1,
    1,
    '2023-01-15'
 ),
 (
    1,
    2,
    '2023-02-01'
 ),
 (
    2,
    1,
    '2023-01-20'
 ),
 (
    3,
    2,
    '2023-03-05'
 ),
 (
    4,
    3,
    '2023-04-01'
 );
 INSERT
    INTO
    amaksatov.lesson_views (
        user_id,
        lesson_id,
        viewed_at
    )
 VALUES
 (
    1,
    1,
    '2023-01-16 10:00:00'
 ),
 (
    1,
    2,
    '2023-01-16 10:15:00'
 ),
 (
    2,
    1,
    '2023-01-21 09:00:00'
 ),
 (
    3,
    3,
    '2023-03-06 12:00:00'
 ),
 (
    4,
    4,
    '2023-04-02 10:00:00'
 ),
 (
    4,
    5,
    '2023-04-03 10:00:00'
 );
@@ -0,0 +1,68 @@
 CREATE MATERIALIZED VIEW user_activity_summary AS
 SELECT
    u.id AS user_id, -- user_id
    u.name, -- name
    COUNT(DISTINCT e.course_id) AS enrollment_count, -- кол-во записей на курсы
    COUNT(DISTINCT lv.lesson_id) AS viewed_lessons_count, -- кол-во просмотренных уроков
    MAX(lv.viewed_at) AS last_viewed, -- последний просмотр
    MIN(lv.viewed_at) AS first_viewed, -- первый просмотр
    ROUND(COUNT(DISTINCT l.course_id) * 100.0 / COUNT(DISTINCT l.id), 2) || '%' AS engagement_rate -- вовлеченность(процент просмотренных/записанным)
 FROM amaksatov.users u -- основная таблица
 JOIN enrollments e ON e.user_id = u.id -- связь users с enrollments
 LEFT JOIN lesson_views lv ON lv.user_id = e.user_id -- связь lessons с lesson_views
 JOIN lessons l ON l.course_id = e.course_id  -- связь уроков с курсами пользователя
 GROUP BY u.id, u.name -- группировка по user_id и name
 ORDER BY user_id; -- сортировка по user_id
 CREATE MATERIALIZED VIEW course_summary AS
 SELECT
    c.id AS course_id,
    c.title AS course_title,
    COUNT(DISTINCT e.user_id) AS unique_users,
    COUNT(DISTINCT l.id) AS lessons_count,
    COUNT(DISTINCT lv.id) AS total_views,
    ROUND(
        CASE
            WHEN COUNT(DISTINCT e.user_id) = 0 THEN NULL
            ELSE COUNT(DISTINCT lv.id) * 1.0 / COUNT(DISTINCT e.user_id)
        END, 2) AS avg_views_per_user,
    MIN(lv.viewed_at) AS first_viewed_at,
    MAX(lv.viewed_at) AS last_viewed_at
 FROM courses c
 LEFT JOIN enrollments e ON e.course_id = c.id
 LEFT JOIN lessons l ON c.id = l.course_id
 LEFT JOIN lesson_views lv ON lv.lesson_id = l.id
 GROUP BY c.id;
 CREATE MATERIALIZED VIEW platform_summary AS
 WITH
 user_stats AS (
    SELECT COUNT(*) AS total_users FROM users
 ),
 course_stats AS (
    SELECT COUNT(*) AS total_courses FROM courses
 ),
 lesson_stats AS (
    SELECT COUNT(*) AS total_lessons FROM lessons
 ),
 view_stats AS (
    SELECT COUNT(*) AS total_views, COUNT(DISTINCT user_id) AS users_with_views
    FROM lesson_views
 )
 SELECT
    u.total_users,
    c.total_courses,
    l.total_lessons,
    v.users_with_views,
    ROUND(l.total_lessons::decimal / NULLIF(c.total_courses, 0), 2) AS avg_lessons_per_course,
    ROUND(v.total_views::decimal / NULLIF(l.total_lessons, 0), 2) AS avg_views_per_lesson
 FROM user_stats u, course_stats c, lesson_stats l, view_stats v;