Merge pull request 'Polikanov63' (#21) from Polikanov63 into main

Reviewed-on: https://gitea.de-infra.servehttp.com/de-infrastructure/DWH/pulls/21
Reviewed-by: aleksey bogatyrev <abogatyrev@example.com>
This commit is contained in:
2025-12-14 06:27:23 +04:00
22 changed files with 691 additions and 0 deletions
@@ -0,0 +1,50 @@
DROP SCHEMA IF EXISTS edu_platform CASCADE;
CREATE SCHEMA edu_platform;
drop table if exists edu_platform.users;
CREATE TABLE edu_platform.users (
id SERIAL PRIMARY KEY,
name VARCHAR(100) NOT NULL,
age INT NOT NULL,
email VARCHAR(100) NOT NULL,
registration_date DATE NOT NULL
);
drop table if exists edu_platform.courses;
CREATE TABLE edu_platform.courses (
id SERIAL PRIMARY KEY,
title VARCHAR(100) NOT NULL,
category VARCHAR(100) NOT NULL,
created_at DATE not NULL
);
drop table if exists edu_platform.lessons;
create TABLE edu_platform.lessons (
id SERIAL PRIMARY KEY,
course_id INT NOT NULL,
title VARCHAR(100) NOT NULL,
duration_min INT NOT null,
FOREIGN KEY (course_id) REFERENCES edu_platform.courses(id)
);
drop table if exists edu_platform.enrollments;
create TABLE edu_platform.enrollments (
id SERIAL PRIMARY KEY,
user_id INT NOT NULL,
course_id INT NOT NULL,
enrolled_at DATE not null,
foreign key (user_id) references edu_platform.users(id),
foreign key (course_id) references edu_platform.courses(id)
);
drop table if exists edu_platform.lesson_views;
create table edu_platform.lesson_views (
id SERIAL PRIMARY KEY,
user_id INT NOT NULL,
lesson_id INT NOT NULL,
viewed_at TIMESTAMP not null,
foreign key (user_id) references edu_platform.users(id),
foreign key (lesson_id) references edu_platform.lessons(id)
);
@@ -0,0 +1,38 @@
INSERT INTO edu_platform.users (id, name, age, email, registration_date) VALUES
(1, 'Alice', 25, 'alice@mail.com', '2023-01-10'),
(2,'Bob', 30, 'bob@gmail.com', '2023-02-05'),
(3,'Charlie', 22, 'charlie@mail.com', '2023-02-20'),
(4,'Diana', 28, 'diana@mail.com', '2023-03-01'),
(5,'Ethan', 35, 'ethan@gmail.com', '2023-03-10');
INSERT INTO edu_platform.courses (id, title, category, created_at) VALUES
(1, 'SQL для начинающих', 'data', '2023-01-01'),
(2, 'Python для анализа данных', 'programming', '2023-01-15'),
(3, 'BI с нуля', 'business', '2023-03-01');
-- Таблица lessons
INSERT INTO edu_platform.lessons (id, course_id, title, duration_min) VALUES
(1, 1, 'SELECT и FROM', 10),
(2, 1, 'JOIN', 15),
(3, 2, 'Pandas', 20),
(4, 3, 'Основы BI', 12),
(5, 3, 'Метрики и дашборды', 18);
-- Таблица enrollments
INSERT INTO edu_platform.enrollments (id, user_id, course_id, enrolled_at) VALUES
(1, 1, 1, '2023-01-15'),
(2, 1, 2, '2023-02-01'),
(3, 2, 1, '2023-01-20'),
(4, 3, 2, '2023-03-05'),
(5, 4, 3, '2023-04-01');
-- Таблица lesson_views
INSERT INTO edu_platform.lesson_views (id, user_id, lesson_id, viewed_at) VALUES
(1, 1, 1, '2023-01-16 10:00:00'),
(2, 1, 2, '2023-01-16 10:15:00'),
(3, 2, 1, '2023-01-21 09:00:00'),
(4, 3, 3, '2023-03-06 12:00:00'),
(5, 4, 4, '2023-04-02 10:00:00'),
(6, 4, 5, '2023-04-03 10:00:00');
@@ -0,0 +1,74 @@
-- Активность пользователей
-- До анализа EXPLAIN:
drop Materialized VIEW if exists edu_platform.user_activity_summary;
create Materialized VIEW edu_platform.user_activity_summary as
select
u.id as user_id ,
u.name as name,
COUNT(distinct e.course_id) as enrollment_count,
COUNT(distinct lv.lesson_id) as viewed_lessons_count,
MAX(lv.viewed_at) as last_view,
MIN(lv.viewed_at) as first_view,
ROUND(
COUNT(DISTINCT l.id) * 100.0 / NULLIF(COUNT(distinct l2.id), 0),
2) as engagement_rate
from
edu_platform.users u
left join edu_platform.enrollments e on u.id = e.user_id
inner join edu_platform.lesson_views lv on u.id = lv.user_id
left join edu_platform.lessons l on lv.lesson_id = l.id
left join edu_platform.lessons l2 on e.course_id = l2.course_id
group by
u.id, u.name;
-- После анализа EXPLAIN
-- Убрал диблирование строк, возникающее при Join-ах.
-- Минимизировал DISTINCT.
drop Materialized VIEW if exists edu_platform.user_activity_summary;
create Materialized VIEW edu_platform.user_activity_summary as
WITH
course_lessons AS (
SELECT course_id, COUNT(*) AS lesson_count
FROM edu_platform.lessons
GROUP BY course_id
),
users_enroll as (
select
u.id as user_id ,
u.name as name,
COUNT(DISTINCT e.course_id) as enrollment_count,
SUM(cl.lesson_count) as lesson_sum
from
edu_platform.users u
left join edu_platform.enrollments e on u.id = e.user_id
left join course_lessons cl on e.course_id = cl.course_id
group by u.id, u.name
),
users_lesson_views as (
select
lv.user_id,
COUNT(distinct lv.lesson_id) as viewed_lessons_count,
MAX(lv.viewed_at) as last_view,
MIN(lv.viewed_at) as first_view
from
edu_platform.lesson_views lv
group by lv.user_id
)
select
ue.user_id,
ue.name,
ue.enrollment_count,
ulv.viewed_lessons_count,
ulv.last_view,
ulv.first_view,
ROUND( (nullif(ulv.viewed_lessons_count, 0) * 100.0) / ue.lesson_sum,
2) as engagement_rate
from users_enroll ue
left join users_lesson_views ulv on ue.user_id = ulv.user_id;
@@ -0,0 +1,51 @@
-- Эффективность курсов
-- После анализа EXPLAIN
-- Убрал диблирование строк, возникающее при Join-ах.
-- Минимизировал DISTINCT.
drop Materialized VIEW if exists edu_platform.course_summary;
create Materialized VIEW edu_platform.course_summary as
with
enroll as (
select
course_id,
COUNT(DISTINCT user_id) as unique_users -- Может ли один и тот же юзер записаться на курс больше одного раза? Если да, то нужен DISTINCT
from
edu_platform.enrollments
group by
course_id),
lessons as (
select
course_id,
COUNT(*) as lessons_count
from
edu_platform.lessons
group by
course_id),
views as (
select
l.course_id as course_id,
count(*) as total_views,
MIN(lv.viewed_at) as first_view,
MAX(lv.viewed_at) as last_view
from
edu_platform.lesson_views lv
left join edu_platform.lessons l on lv.lesson_id = l.id
group by
l.course_id)
select
c.id AS course_id,
c.title AS course_title,
e.unique_users,
l.lessons_count,
v.total_views,
ROUND( v.total_views * 1.0 / nullif(e.unique_users, 0)
,2
) as avg_views_per_user,
v.first_view,
v.last_view
from
edu_platform.courses c
LEFT JOIN enroll e ON c.id = e.course_id
LEFT JOIN lessons l ON c.id = l.course_id
LEFT JOIN views v ON c.id = v.course_id;
@@ -0,0 +1,21 @@
-- Общая отчетность по взаимодействию
drop Materialized VIEW if exists edu_platform.platform_summary;
create Materialized VIEW edu_platform.platform_summary as
select
(select COUNT(*) from edu_platform.users) as total_users,
(select COUNT(*) from edu_platform.courses) as total_courses,
(select COUNT(*) from edu_platform.lessons) as total_lessons,
(select COUNT(distinct user_id) from edu_platform.lesson_views) as users_with_views,
ROUND(
(select AVG(t1.cnt) from
(select course_id, COUNT(id) cnt
from edu_platform.lessons
group by course_id) t1)
,2) as avg_lessons_per_course,
ROUND(
(select AVG(t2.cnt) from
(select lesson_id, COUNT(id) cnt
from edu_platform.lesson_views
group by lesson_id) t2)
,2) as avg_views_per_lesson;
@@ -0,0 +1,64 @@
-- 0. Создаём БД
CREATE DATABASE dwh_db
WITH OWNER = tester
TEMPLATE = template0
ENCODING = 'UTF8';
-- 1. Создём схемы
CREATE SCHEMA raw; -- сырые данные
CREATE SCHEMA core; -- или core. Тут хранилище: факты и измерения
CREATE SCHEMA dm; -- BI-витрины
-- 2. Создаём raw-таблицы:
CREATE TABLE raw.users (
id serial4 NOT NULL,
"name" varchar(100) NOT NULL,
age int4 NOT NULL,
email varchar(100) NOT NULL,
registration_date date NOT NULL,
CONSTRAINT users_pkey PRIMARY KEY (id)
)
DISTRIBUTED BY (id);
CREATE TABLE raw.courses (
id serial4 NOT NULL,
title varchar(100) NOT NULL,
category varchar(100) NOT NULL,
created_at date NOT NULL,
CONSTRAINT courses_pkey PRIMARY KEY (id)
)
DISTRIBUTED BY (id);
CREATE TABLE raw.lessons (
id serial4 NOT NULL,
course_id int4 NOT NULL,
title varchar(100) NOT NULL,
duration_min int4 NOT NULL,
CONSTRAINT lessons_pkey PRIMARY KEY (id)
)
DISTRIBUTED BY (id);
ALTER TABLE raw.lessons ADD CONSTRAINT lessons_course_id_fkey FOREIGN KEY (course_id) REFERENCES raw.courses(id);
CREATE TABLE raw.enrollments (
id serial4 NOT NULL,
user_id int4 NOT NULL,
course_id int4 NOT NULL,
enrolled_at date NOT NULL,
CONSTRAINT enrollments_pkey PRIMARY KEY (id)
)
DISTRIBUTED BY (id);
ALTER TABLE raw.enrollments ADD CONSTRAINT enrollments_course_id_fkey FOREIGN KEY (course_id) REFERENCES raw.courses(id);
ALTER TABLE raw.enrollments ADD CONSTRAINT enrollments_user_id_fkey FOREIGN KEY (user_id) REFERENCES raw.users(id);
CREATE TABLE raw.lesson_views (
id serial4 NOT NULL,
user_id int4 NOT NULL,
lesson_id int4 NOT NULL,
viewed_at timestamp NOT NULL,
CONSTRAINT lesson_views_pkey PRIMARY KEY (id)
)
DISTRIBUTED BY (id);
ALTER TABLE raw.lesson_views ADD CONSTRAINT lesson_views_lesson_id_fkey FOREIGN KEY (lesson_id) REFERENCES raw.lessons(id);
ALTER TABLE raw.lesson_views ADD CONSTRAINT lesson_views_user_id_fkey FOREIGN KEY (user_id) REFERENCES raw.users(id);
@@ -0,0 +1,48 @@
/* Копирую в файлы в контейнер:
docker cp D:/work/code/DE/GP/Pet-project/IMPORT_DATA/users_202512050730.csv gpdb1:/tmp/users_202512050730.csv
docker cp D:/work/code/DE/GP/Pet-project/IMPORT_DATA/courses_202512050730.csv gpdb1:/tmp/courses_202512050730.csv
docker cp D:/work/code/DE/GP/Pet-project/IMPORT_DATA/enrollments_202512050730.csv gpdb1:/tmp/enrollments_202512050730.csv
docker cp D:/work/code/DE/GP/Pet-project/IMPORT_DATA/lesson_views_202512050730.csv gpdb1:/tmp/lesson_views_202512050730.csv
docker cp D:/work/code/DE/GP/Pet-project/IMPORT_DATA/lessons_202512050730.csv gpdb1:/tmp/lessons_202512050730.csv
*/
COPY raw.users
FROM '/tmp/users_202512050730.csv'
WITH (
FORMAT csv,
HEADER true,
DELIMITER ','
);
COPY raw.courses
FROM '/tmp/courses_202512050730.csv'
WITH (
FORMAT csv,
HEADER true,
DELIMITER ','
);
COPY raw.lessons
FROM '/tmp/lessons_202512050730.csv'
WITH (
FORMAT csv,
HEADER true,
DELIMITER ','
);
COPY raw.enrollments
FROM '/tmp/enrollments_202512050730.csv'
WITH (
FORMAT csv,
HEADER true,
DELIMITER ','
);
COPY raw.lesson_views
FROM '/tmp/lesson_views_202512050730.csv'
WITH (
FORMAT csv,
HEADER true,
DELIMITER ','
);
@@ -0,0 +1,51 @@
-- Измерение пользователей
CREATE TABLE core.dim_user (
user_id INT PRIMARY KEY,
name TEXT,
age INT,
email TEXT,
registration_date DATE
)
DISTRIBUTED BY (user_id);
-- Измерение курсов
CREATE TABLE core.dim_course (
course_id INT PRIMARY KEY,
title TEXT,
category TEXT
)
DISTRIBUTED BY (course_id);
-- Измерение уроков
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 BY (lesson_id);
-- Опционально: измерение времени
CREATE TABLE core.dim_date (
date DATE PRIMARY KEY,
year INT,
month INT,
day INT,
weekday TEXT
)
DISTRIBUTED BY (date);
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);
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);
@@ -0,0 +1,49 @@
BEGIN;
-- UPDATE с проверкой изменений
update
core.dim_user d
set
name = s.name,
age = s.age,
email = s.email,
registration_date = s.registration_date
from
(
select
distinct
id::INT,
name::TEXT,
age::INT,
email::TEXT,
registration_date::DATE
from
raw.users
) s
where
d.user_id = s.id
and (
d.name is distinct from s.name
or d.age is distinct from s.age
or d.email is distinct from s.email
or d.registration_date is distinct from s.registration_date
);
-- INSERT новых строк
INSERT INTO core.dim_user (user_id, name, age, email, registration_date)
SELECT
s.id,
s.name,
s.age,
s.email,
s.registration_date
FROM (
SELECT DISTINCT
id::INT,
name::TEXT,
age::INT,
email::TEXT,
registration_date::DATE
FROM raw.users
) s
LEFT JOIN core.dim_user d on d.user_id = s.id
WHERE d.user_id IS NULL;
COMMIT;
@@ -0,0 +1,39 @@
BEGIN;
-- UPDATE с проверкой изменений
update
core.dim_course dc
set
title = s.title,
category = s.category
from
(
select
distinct
id::INT,
title::TEXT,
category::TEXT
from
raw.courses
) s
where
dc.course_id = s.id
and (
dc.title is distinct from s.title
or dc.category is distinct from s.category
);
-- INSERT новых строк
INSERT INTO core.dim_course (course_id, title, category)
SELECT
s.id,
s.title,
s.category
FROM (
SELECT DISTINCT
id::INT,
title::TEXT,
category::TEXT
FROM raw.courses
) s
LEFT JOIN core.dim_course cd on cd.course_id = s.id
WHERE cd.course_id IS NULL;
COMMIT;
@@ -0,0 +1,44 @@
BEGIN;
-- UPDATE с проверкой изменений
update
core.dim_lesson dl
set
title = l.title,
duration_min = l.duration_min,
course_id = l.course_id
from
(
select distinct
id::INT,
course_id::INT,
title::TEXT,
duration_min::INT
from
raw.lessons
) l
where
dl.lesson_id = l.id
and (
dl.title is distinct from l.title
or dl.duration_min is distinct from l.duration_min
or dl.course_id is distinct from l.course_id
);
-- INSERT новых строк
INSERT INTO core.dim_lesson (lesson_id, title, duration_min, course_id)
SELECT
l.id,
l.title,
l.duration_min,
l.course_id
FROM (
select distinct
id::INT,
course_id::INT,
title::TEXT,
duration_min::INT
from
raw.lessons
) l
LEFT JOIN core.dim_lesson dl on dl.lesson_id = l.id
WHERE dl.lesson_id IS NULL;
COMMIT;
@@ -0,0 +1,17 @@
insert into core.enrollments (user_id, course_id, enrolled_at)
select
re.user_id,
re.course_id,
re.enrolled_at
from
(select DISTINCT
user_id::INT,
course_id::INT,
enrolled_at
from
raw.enrollments) re
LEFT join core.enrollments ce on
ce.user_id = re.user_id
and ce.course_id = re.course_id
and ce.enrolled_at = re.enrolled_at
WHERE ce.user_id IS NULL;
@@ -0,0 +1,22 @@
-- INSERT новых строк
INSERT INTO core.fact_lesson_views (user_id, lesson_id, course_id, viewed_at)
select
l.user_id,
l.lesson_id,
l.course_id,
l.viewed_at
from (select
lv.user_id,
lv.lesson_id,
l.course_id,
lv.viewed_at
from
raw.lesson_views lv
left join raw.lessons l on lv.lesson_id = l.id
) l
LEFT join core.fact_lesson_views flv on
flv.user_id = l.user_id
and flv.lesson_id = l.lesson_id
and flv.course_id = l.course_id
and flv.viewed_at = l.viewed_at
WHERE flv.user_id IS NULL;
@@ -0,0 +1,29 @@
DROP MATERIALIZED VIEW IF EXISTS dm.lesson_popularity_summary;
CREATE MATERIALIZED VIEW dm.lesson_popularity_summary
DISTRIBUTED BY (lesson_id)
AS
select
t1.lesson_id as lesson_id,
dl.title as lesson_title,
t1.course_id as course_id,
dc.title as course_title,
t1.total_views as total_views,
t1.unique_users as unique_users,
t1.first_view as first_view,
t1.last_view as last_view
from
(select
flv.lesson_id,
flv.course_id,
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.fact_lesson_views flv
group by
flv.lesson_id,
flv.course_id
) as t1
right join core.dim_lesson dl on dl.lesson_id = t1.lesson_id
left join core.dim_course dc on t1.course_id = dc.course_id;
@@ -0,0 +1,25 @@
DROP MATERIALIZED VIEW IF EXISTS dm.inactive_users_summary;
CREATE MATERIALIZED VIEW dm.inactive_users_summary
DISTRIBUTED BY (user_id)
AS
select
du.user_id as user_id,
du.name as user_name,
du.email as email,
du.age as age,
COALESCE(e.enrollments_count, 0) as enrollments_count,
du.registration_date as registration_date
from
core.dim_user du
left join
(select
user_id,
COUNT(*) as enrollments_count
from
core.enrollments
group by
user_id
) e on du.user_id = e.user_id
left join core.fact_lesson_views flv on du.user_id = flv.user_id
where
flv.user_id is null;
@@ -0,0 +1,39 @@
DROP MATERIALIZED VIEW IF EXISTS dm.course_completion_rate;
CREATE MATERIALIZED VIEW dm.course_completion_rate
DISTRIBUTED BY (user_id)
AS
with activity_users as (
select
e.user_id as user_id,
e.course_id as course_id,
COUNT(DISTINCT flv.viewed_at) as lessons_viewed
from
core.enrollments e
left join core.dim_lesson dl on e.course_id = dl.course_id
left join core.fact_lesson_views flv on e.user_id = flv.user_id and e.course_id = flv.course_id and dl.lesson_id = flv.lesson_id
group by
e.user_id,
e.course_id),
course2 as (
select
course_id,
COUNT(*) as lessons_in_course
from core.dim_lesson
group by
course_id)
select
au.user_id as user_id,
du.name as user_name,
au.course_id as course_id,
dc.title as course_title,
c2.lessons_in_course as lessons_in_course,
au.lessons_viewed as lessons_viewed,
ROUND(au.lessons_viewed * 1.0 / c2.lessons_in_course , 2) as completion_rate
from
activity_users au
left join core.dim_user du on au.user_id = du.user_id
left join core.dim_course dc on au.course_id = dc.course_id
left join course2 c2 on au.course_id = c2.course_id
order by
au.user_id,
au.course_id;
@@ -0,0 +1,4 @@
"id","title","category","created_at"
1,SQL для начинающих,data,2023-01-01
2,Python для анализа данных,programming,2023-01-15
3,BI с нуля,business,2023-03-01
1 id title category created_at
2 1 SQL для начинающих data 2023-01-01
3 2 Python для анализа данных programming 2023-01-15
4 3 BI с нуля business 2023-03-01
@@ -0,0 +1,6 @@
"id","user_id","course_id","enrolled_at"
1,1,1,2023-01-15
2,1,2,2023-02-01
3,2,1,2023-01-20
4,3,2,2023-03-05
5,4,3,2023-04-01
1 id user_id course_id enrolled_at
2 1 1 1 2023-01-15
3 2 1 2 2023-02-01
4 3 2 1 2023-01-20
5 4 3 2 2023-03-05
6 5 4 3 2023-04-01
@@ -0,0 +1,7 @@
"id","user_id","lesson_id","viewed_at"
1,1,1,2023-01-16 10:00:00.000
2,1,2,2023-01-16 10:15:00.000
3,2,1,2023-01-21 09:00:00.000
4,3,3,2023-03-06 12:00:00.000
5,4,4,2023-04-02 10:00:00.000
6,4,5,2023-04-03 10:00:00.000
1 id user_id lesson_id viewed_at
2 1 1 1 2023-01-16 10:00:00.000
3 2 1 2 2023-01-16 10:15:00.000
4 3 2 1 2023-01-21 09:00:00.000
5 4 3 3 2023-03-06 12:00:00.000
6 5 4 4 2023-04-02 10:00:00.000
7 6 4 5 2023-04-03 10:00:00.000
@@ -0,0 +1,6 @@
"id","course_id","title","duration_min"
1,1,SELECT и FROM,10
2,1,JOIN,15
3,2,Pandas,20
4,3,Основы BI,12
5,3,Метрики и дашборды,18
1 id course_id title duration_min
2 1 1 SELECT и FROM 10
3 2 1 JOIN 15
4 3 2 Pandas 20
5 4 3 Основы BI 12
6 5 3 Метрики и дашборды 18
@@ -0,0 +1,6 @@
"id","name","age","email","registration_date"
1,Alice,25,alice@mail.com,2023-01-10
2,Bob,30,bob@gmail.com,2023-02-05
3,Charlie,22,charlie@mail.com,2023-02-20
4,Diana,28,diana@mail.com,2023-03-01
5,Ethan,35,ethan@gmail.com,2023-03-10
1 id name age email registration_date
2 1 Alice 25 alice@mail.com 2023-01-10
3 2 Bob 30 bob@gmail.com 2023-02-05
4 3 Charlie 22 charlie@mail.com 2023-02-20
5 4 Diana 28 diana@mail.com 2023-03-01
6 5 Ethan 35 ethan@gmail.com 2023-03-10
+1
View File
@@ -0,0 +1 @@
test коммита в Gitea