ddmitry/de-roadmap
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

переписана реализация scd2

Browse Source
This commit is contained in:
ddadmin
2025-12-21 18:50:50 +03:00
parent 87301921a9
commit ae00475df5
15 changed files with 205 additions and 183 deletions
Download Patch File Download Diff File Expand all files Collapse all files
dwh-modeling/README.md
+20 -19
View File
@@ -234,13 +234,12 @@ erDiagram
dim_customer {
bigint customer_sk PK "суррогатный ключ"
varchar customer_bk "бизнес-ключ, напр. '101'"
int customer_bk "бизнес-ключ, напр. 101"
varchar customer_name
varchar email
varchar city
date valid_from "SCD2: с какой даты запись актуальна"
date valid_to "SCD2: по какую дату актуальна"
boolean is_current "SCD2: текущая версия?"
date valid_to "SCD2: по какую дату актуальна (NULL = сейчас)"
}
dim_product {
@@ -269,15 +268,16 @@ erDiagram
В `dim_customer` это будет **три строки**:
| customer_sk | customer_bk | email | city | valid_from | valid_to | is_current |
|-------------|-------------|-------|------|------------|----------|------------|
| 1001 | 101 | a@ex.com | Москва | 2023-01-01 | 2023-05-15 | false |
| 1002 | 101 | b@ex.com | Москва | 2023-05-16 | 2023-09-30 | false |
| 1003 | 101 | b@ex.com | СПб | 2023-10-01 | 9999-12-31 | true |
| customer_sk | customer_bk | email | city | valid_from | valid_to |
|-------------|-------------|-------|------|------------|----------|
| 1001 | 101 | a@ex.com | Москва | 2023-01-01 | 2023-05-16 |
| 1002 | 101 | b@ex.com | Москва | 2023-05-16 | 2023-10-01 |
| 1003 | 101 | b@ex.com | СПб | 2023-10-01 | NULL |
Когда мы считаем продажи за **12 января** — джойним `fact_sales` к той строке `dim_customer`, где:
```sql
fact_sales.order_date BETWEEN dim_customer.valid_from AND dim_customer.valid_to
fact_sales.order_date >= dim_customer.valid_from
AND (dim_customer.valid_to IS NULL OR fact_sales.order_date < dim_customer.valid_to)
```
и получаем актуальный на тот день email и город.
@@ -554,8 +554,8 @@ JOIN dds.dim_product p
ON f.product_sk = p.product_sk
JOIN dds.dim_customer c
ON f.customer_sk = c.customer_sk
AND f.order_date BETWEEN c.valid_from AND c.valid_to -- SCD!
WHERE c.is_current = true -- или не фильтровать — тогда будет история
AND f.order_date >= c.valid_from
AND (c.valid_to IS NULL OR f.order_date < c.valid_to) -- SCD!
GROUP BY d.date_actual, p.product_name, c.customer_segment;
```
@@ -775,10 +775,11 @@ SELECT 'OK' WHERE EXISTS (
[`customers.csv`](data/customers.csv):
```csv
customer_id,email,phone,city
101,a@ex.com,700,Москва
101,b@ex.com,700,Москва
102,c@ex.com,701,СПб
customer_id,email,phone,city,event_ts,_load_id,load_ts
101,a@ex.com,700,Москва,2024-01-01,batch_20240101_0800,2024-01-01 08:00
102,c@ex.com,701,СПб,2024-01-01,batch_20240101_0800,2024-01-01 08:00
101,b@ex.com,700,Москва,2024-05-16,batch_20240516_0800,2024-05-16 08:00
101,b@ex.com,700,Санкт-Петербург,2024-10-01,batch_20241001_0800,2024-10-01 08:00
```
[`orders.csv`](data/orders.csv):
@@ -807,7 +808,8 @@ product_id,name
```csv
product_id,valid_from,valid_to,price
9001,2023-12-01,2024-01-31,100
9001,2024-02-01,2999-12-31,110
9001,2024-02-01,,110
9002,2023-01-01,,50
```
> 📂 Все SQL-скрипты для построения хранилища находятся в папке [`sql/`](sql/).
@@ -824,13 +826,12 @@ product_id,valid_from,valid_to,price
-- DDS: измерение клиента (SCD Type 2)
CREATE TABLE dds.dim_customer (
customer_sk BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
customer_bk VARCHAR(50) NOT NULL, -- напр. '101'
customer_bk INT NOT NULL, -- напр. 101
email VARCHAR(100),
phone VARCHAR(20),
city VARCHAR(50),
valid_from DATE NOT NULL,
valid_to DATE DEFAULT '9999-12-31',
is_current BOOLEAN NOT NULL DEFAULT TRUE
valid_to DATE
);
-- DDS: факт продаж (гранулярность: строка заказа)