Files
homework/bhabibullin/greenplum/pet_project_gp.sql
T

218 lines
5.8 KiB
SQL

CREATE SCHEMA raw;
CREATE SCHEMA core;
CREATE SCHEMA dm;
CREATE TABLE raw.users (
id INT,
name VARCHAR(255),
age INT,
email VARCHAR(255),
registration_date DATE
);
CREATE TABLE raw.courses (
id INT,
title VARCHAR(255),
category VARCHAR(255),
created_at DATE
);
CREATE TABLE raw.lessons (
id INT,
course_id INT,
title VARCHAR(255),
duration_min INT
);
CREATE TABLE raw.enrollments (
id INT,
user_id INT,
course_id INT,
enrolled_at DATE
);
CREATE TABLE raw.lesson_views (
id INT,
user_id INT,
lesson_id INT,
viewed_at TIMESTAMP
);
CREATE TABLE core.dim_user (
user_id INT PRIMARY KEY,
name TEXT,
age INT,
email TEXT,
registration_date DATE
);
CREATE TABLE core.dim_course (
course_id INT PRIMARY KEY,
title TEXT,
category TEXT
);
CREATE TABLE core.dim_lesson (
lesson_id INT PRIMARY KEY,
title TEXT,
duration_min INT,
course_id INT REFERENCES core.dim_course(course_id)
);
CREATE TABLE core.fact_lesson_views (
user_id INT REFERENCES core.dim_user(user_id),
lesson_id INT REFERENCES core.dim_lesson(lesson_id),
course_id INT REFERENCES core.dim_course(course_id),
viewed_at TIMESTAMP
);
CREATE TABLE core.enrollments (
user_id INT REFERENCES core.dim_user(user_id),
course_id INT REFERENCES core.dim_course(course_id),
enrolled_at TIMESTAMP
);
INSERT INTO core.dim_user (user_id, name, age, email, registration_date)
SELECT DISTINCT
id::INT,
name::TEXT,
age::INT,
email::TEXT,
registration_date::DATE
FROM raw.users;
INSERT INTO core.dim_course (course_id, title, category)
SELECT DISTINCT
id::INT,
title::TEXT,
category::TEXT
FROM raw.courses;
INSERT INTO core.dim_lesson (lesson_id, title, duration_min, course_id)
SELECT DISTINCT
id::INT,
title::TEXT,
duration_min::INT,
course_id::INT
FROM raw.lessons;
INSERT INTO core.fact_lesson_views (user_id, lesson_id, course_id, viewed_at)
SELECT DISTINCT
rw.user_id::INT,
rw.lesson_id::INT,
l.course_id::INT,
rw.viewed_at::TIMESTAMP
FROM raw.lesson_views rw
JOIN raw.lessons l ON rw.lesson_id = l.id;
INSERT INTO core.enrollments (user_id, course_id, enrolled_at)
SELECT DISTINCT
user_id::INT,
course_id::INT,
enrolled_at::DATE
FROM raw.enrollments;
CREATE MATERIALIZED VIEW dm.dm_lesson_popularity_summary AS
SELECT
dl.lesson_id,
dl.title AS lesson_title,
dc.course_id,
dc.title AS course_title,
COUNT(*) AS total_views,
COUNT(DISTINCT flv.user_id) AS unique_users,
MIN(flv.viewed_at) AS first_view,
MAX(flv.viewed_at) AS last_view
FROM core.dim_lesson dl
JOIN core.dim_course dc ON dc.course_id = dl.course_id
JOIN core.fact_lesson_views flv ON flv.lesson_id = dl.lesson_id
GROUP BY dl.lesson_id, dl.title, dc.course_id, dc.title;
/*
Цель:
Показать популярность каждого урока.
Поля витрины:
- lesson_id — идентификатор урока
- lesson_title — название урока
- course_id — идентификатор курса
- course_title — название курса
- total_views — общее число просмотров урока
- unique_users — количество уникальных пользователей
- first_view — дата первого просмотра
- last_view — дата последнего просмотра
*/
CREATE MATERIALIZED VIEW dm.dm_inactive_users_summary AS
SELECT
du.user_id,
du.name AS user_name,
du.email,
du.age,
COUNT(DISTINCT e.course_id) AS enrollments_count,
du.registration_date
FROM core.dim_user du
LEFT JOIN core.enrollments e ON e.user_id = du.user_id
WHERE NOT EXISTS (
SELECT 1
FROM core.fact_lesson_views flv
WHERE flv.user_id = du.user_id
)
GROUP BY du.user_id, du.name, du.email, du.age, du.registration_date;
/*
Цель:
Определить невовлечённых пользователей и потенциальный отток.
Поля витрины:
- user_id — идентификатор пользователя
- user_name — имя
- email — почта
- age — возраст
- enrollments_count — количество курсов
- registration_date — дата регистрации
*/
CREATE MATERIALIZED VIEW dm.course_completion_rate AS
SELECT
du.user_id,
du.name AS user_name,
c.course_id,
c.title AS course_title,
COUNT(DISTINCT dl.lesson_id) AS lessons_in_course,
COUNT(DISTINCT flv.lesson_id) AS lessons_viewed,
ROUND(
COUNT(DISTINCT flv.lesson_id) * 1.0 /
NULLIF(COUNT(DISTINCT dl.lesson_id), 0),
2
) AS completion_rate
FROM core.dim_user du
JOIN core.enrollments e ON du.user_id = e.user_id
JOIN core.dim_course c ON e.course_id = c.course_id
JOIN core.dim_lesson dl ON c.course_id = dl.course_id
LEFT JOIN core.fact_lesson_views flv ON dl.lesson_id = flv.lesson_id
AND flv.user_id = du.user_id
GROUP BY du.user_id, du.name, c.course_id, c.title
ORDER BY du.user_id, c.course_id;
/*
Цель:
Анализировать, насколько пользователи проходят курсы до конца.
Поля витрины:
- user_id — идентификатор пользователя
- user_name — имя пользователя
- course_id — идентификатор курса
- course_title — название курса
- total_lessons — количество уроков в курсе
- lessons_viewed — число просмотренных уроков
- completion_rate — lessons_viewed / total_lessons
*/