Merge pull request 'added STEP 1, 3, 4.3' (#32) from OS-233 into main

Reviewed-on: https://gitea.de-infra.servehttp.com/de-infrastructure/DWH/pulls/32
Reviewed-by: aleksey bogatyrev <abogatyrev@example.com>
This commit is contained in:
2025-12-29 16:57:30 +04:00
7 changed files with 257 additions and 0 deletions
@@ -0,0 +1,5 @@
--- создаем необх. схемы для DWH
CREATE SCHEMA raw;
CREATE SCHEMA core;
CREATE SCHEMA dm;
@@ -0,0 +1,44 @@
--создаем таблицы для raw слоя с DISTRIBUTED REPLICATED, т.к. таблицы довольно маленькие
CREATE TABLE raw.users (
id BIGINT PRIMARY KEY,
name VARCHAR(100),
age INT,
email VARCHAR(255) UNIQUE,
registration_date DATE NOT NULL
)
DISTRIBUTED REPLICATED;
CREATE TABLE raw.courses(
id BIGINT PRIMARY KEY,
title TEXT NOT NULL,
category VARCHAR(255) CHECK (category IN ('data', 'programming', 'business')),
created_at DATE NOT NULL
)
DISTRIBUTED REPLICATED;
CREATE TABLE raw.lessons(
id BIGINT PRIMARY KEY,
course_id INT NOT NULL,
title TEXT NOT NULL,
duration_min INT NOT NULL
)
DISTRIBUTED REPLICATED;
CREATE TABLE raw.enrollments(
id BIGINT PRIMARY KEY,
user_id INT NOT NULL,
course_id INT NOT NULL,
enrolled_at DATE NOT NULL
)
DISTRIBUTED REPLICATED;
CREATE TABLE raw.lesson_views(
id BIGINT PRIMARY KEY,
user_id INT NOT NULL,
lesson_id INT NOT NULL,
viewed_at TIMESTAMP NOT NULL
)
DISTRIBUTED REPLICATED;
@@ -0,0 +1,95 @@
--- ИЗМЕРЕНИЕ ПОЛЬЗОВАТЕЛЕЙ, РАСПРЕДЕЛЕНИЕ ПО user_id - на проде табл. может быть большой
CREATE TABLE core.dim_user (
user_id INT PRIMARY KEY,
name TEXT,
age INT,
email TEXT,
registration_date DATE
)
DISTRIBUTED BY (user_id);
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;
--- ИЗМЕРЕНИЕ КУРСОВ, РАСПРЕДЕЛЕНИЕ С КОПИРОВАНИЕМ НА НОДУ, т.к. таблица-справочник
CREATE TABLE core.dim_course (
course_id INT PRIMARY KEY,
title TEXT,
category TEXT
)
DISTRIBUTED REPLICATED;
INSERT INTO core.dim_course (course_id, title, category)
SELECT DISTINCT
id::BIGINT,
title::TEXT,
category::VARCHAR(255)
FROM raw.courses;
--- ИЗМЕРЕНИЕ УРОКОВ, РАСПРЕДЕЛЕНИЕ С КОПИРОВАНИЕМ НА НОДУ, т.к. таблица-справочник
CREATE TABLE core.dim_lesson (
lesson_id INT PRIMARY KEY,
title TEXT,
duration_min INT,
course_id INT REFERENCES core.dim_course(course_id)
)
DISTRIBUTED REPLICATED;
INSERT INTO core.dim_lesson (lesson_id, title, duration_min, course_id)
SELECT DISTINCT
id::BIGINT,
title::TEXT,
duration_min::INT,
course_id::INT
FROM raw.lessons;
---- ФАКТ ПРОСМОТРОВ УРОКОВ, РАСПРЕДЕЛЕНИЕ ПО USER_ID ДЛЯ ИСКЛЮЧЕНИЯ MOTION ПРИ JOIN
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
)
DISTRIBUTED BY (user_id);
INSERT INTO core.fact_lesson_views (user_id, lesson_id, course_id, viewed_at)
SELECT
lv.user_id::INT,
lv.lesson_id::INT,
l.course_id::INT,
lv.viewed_at::TIMESTAMP
FROM raw.lesson_views lv
LEFT JOIN raw.lessons l ON lv.lesson_id = l.id;
---- ФАКТ ЗАПИСИ НА КУРСЫ, РАСПРЕДЕЛЕНИЕ ПО USER_ID ДЛЯ ИСКЛЮЧЕНИЯ MOTION ПРИ JOIN
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
)
DISTRIBUTED BY (user_id);
INSERT INTO core.enrollments (user_id, course_id, enrolled_at)
SELECT
user_id::INT,
course_id::INT,
enrolled_at::DATE
FROM raw.enrollments
@@ -0,0 +1,42 @@
/*
- ВИТРИНА АНАЛИЗА ПОПУЛЯРНОСТИ КАЖДОГО УРОКА
- ВСЕ УЧАСТВУЮЩИЕ ИЗМЕРЕНИЯ DISTRIBUTED REPLICATED
- ИТОГОВОЕ ПРЕДСТВАЛЕНИЕ DISTRIBUTED BY (lesson_id, course_id)
*/
CREATE MATERIALIZED VIEW IF NOT EXISTS dm.lesson_popularity_summary AS
-- EXPLAIN ANALYZE
DISTRIBUTED BY (lesson_id, course_id)
WITH t1 AS (
SELECT
lesson_id,
course_id,
COUNT(*) AS total_views,
COUNT(DISTINCT user_id) AS unique_users,
MIN(viewed_at)AS first_view,
MAX(viewed_at)AS last_view
FROM core.fact_lesson_views
GROUP BY lesson_id, course_id
),
t2 AS (
SELECT
l.lesson_id,
l.title AS lesson_title,
c.course_id,
c.title AS course_title
FROM core.dim_course c
LEFT JOIN core.dim_lesson l USING(course_id)
)
SELECT
t1.lesson_id,
t2.lesson_title,
t1.course_id,
t2.course_title,
t1.total_views,
t1.unique_users,
t1.first_view,
t1.last_view
FROM t1
LEFT JOIN t2 USING(lesson_id, course_id);
@@ -0,0 +1,28 @@
/*
- ВИТРИНА АНАЛИЗА НЕАКТИВНЫХ ПОЛЬЗОВАТЕЛЕЙ
- ВСЕ УЧАСТВУЮЩИЕ ТАБЛИЦЫ DISTRIBUTED BY (user_id)
- ИТОГОВОЕ ПРЕДСТВАЛЕНИЕ ТАК ЖЕ DISTRIBUTED BY (user_id)
*/
CREATE MATERIALIZED VIEW IF NOT EXISTS dm.inactive_users_summary AS
-- EXPLAIN ANALYZE
DISTRIBUTED BY (user_id)
WITH t1 AS (
SELECT
user_id,
count(DISTINCT course_id) AS enrollments_count
FROM core.enrollments
GROUP BY user_id
)
SELECT
u.user_id,
u.name AS user_name,
u.email,
u.age,
coalesce(t1.enrollments_count, 0) AS enrollments_count,
u.registration_date
FROM core.dim_user u
LEFT JOIN t1 USING (user_id)
WHERE u.user_id NOT EXISTS (SELECT user_id FROM core.fact_lesson_views);
@@ -0,0 +1,40 @@
/*
- ВИТРИНА ДЛЯ АНАЛИЗА КОНВЕРСИИ ПРОХОЖДЕНИЯ КУРСА ПОЛЬЗОВАТЕЛЯМИ
- CORE.DIM_USER И CORE.FACT_LESSON_VIEWS DISTRIBUTED BY (user_id)
- ИТОГОВОЕ РАСПРЕДЕЛЕНИЕ ПО (user_id, course_id)
*/
CREATE MATERIALIZED VIEW IF NOT EXISTS dm.course_completion_rate AS
WITH t1 AS (
SELECT
e.user_id,
du.name AS user_name,
e.course_id,
dc.title AS course_title,
COUNT(DISTINCT dl.lesson_id) AS lessons_in_course
FROM core.enrollments e
JOIN core.dim_user du USING (user_id)
JOIN core.dim_course dc USING (course_id)
LEFT JOIN core.dim_lesson dl USING (course_id)
GROUP BY user_id, du.name, course_id, dc.title
),
t2 AS (
SELECT
user_id,
course_id,
COUNT(DISTINCT lesson_id) AS lessons_viewed
FROM core.fact_lesson_views
GROUP BY user_id, course_id
)
SELECT
t1.user_id,
t1.user_name,
t1.course_id,
t1.course_title,
t1.lessons_in_course,
COALESCE(t2.lessons_viewed, 0) AS lessons_viewed,
COALESCE(ROUND(lessons_viewed::DECIMAL / NULLIF(lessons_in_course, 0), 2), 0) AS completion_rate
FROM t1
LEFT JOIN t2 USING (user_id, course_id)
DISTRIBUTED BY (user_id, course_id);
+3
View File
@@ -0,0 +1,3 @@
SELECT * FROM dm.lesson_popularity_summary LIMIT 10;
SELECT * FROM dm.inactive_users_summary LIMIT 10;
SELECT * FROM dm.course_completion_rate LIMIT 10;