팩트 테이블과 차원 테이블

데이터 웨어하우스란 무엇인가

팩트 테이블이란

CREATE TABLE fact_sales (
  sale_id      SERIAL PRIMARY KEY,
  date_key     INT NOT NULL,
  product_key  INT NOT NULL,
  customer_key INT NOT NULL,
  store_key    INT NOT NULL,
  quantity     INT NOT NULL,
  unit_price   NUMERIC(10, 2) NOT NULL,
  total_amount NUMERIC(12, 2) NOT NULL
);

차원 테이블이란

CREATE TABLE dim_product (
  product_key  SERIAL PRIMARY KEY,
  product_name VARCHAR(200) NOT NULL,
  category     VARCHAR(100),
  brand        VARCHAR(100),
  unit_cost    NUMERIC(10, 2)
);

CREATE TABLE dim_customer (
  customer_key SERIAL PRIMARY KEY,
  full_name    VARCHAR(200) NOT NULL,
  email        VARCHAR(200),
  country      VARCHAR(100),
  segment      VARCHAR(50)
);

날짜 차원

CREATE TABLE dim_date (
  date_key       INT PRIMARY KEY,  -- e.g. 20240315
  full_date      DATE NOT NULL,
  day_of_week    VARCHAR(10),
  day_of_month   INT,
  month_num      INT,
  month_name     VARCHAR(20),
  quarter        INT,
  year           INT,
  is_holiday     BOOLEAN DEFAULT FALSE,
  fiscal_quarter INT
);

-- Sample row
INSERT INTO dim_date VALUES
  (20240315, '2024-03-15', 'Friday', 15, 3, 'March', 1, 2024, FALSE, 2);

스타 스키마 패턴

-- Join fact to two dimensions to enrich a sales report
SELECT
  dp.product_name,
  dp.category,
  SUM(fs.quantity)     AS total_units_sold,
  SUM(fs.total_amount) AS total_revenue
FROM fact_sales fs
JOIN dim_product  dp ON dp.product_key  = fs.product_key
JOIN dim_date     dd ON dd.date_key     = fs.date_key
WHERE dd.year = 2024
GROUP BY dp.product_name, dp.category
ORDER BY total_revenue DESC;

대체 키와 자연 키 비교

-- Surrogate key approach: integer join is fast
SELECT fs.sale_id, dc.full_name, fs.total_amount
FROM fact_sales fs
JOIN dim_customer dc ON dc.customer_key = fs.customer_key
WHERE dc.country = 'Germany'
LIMIT 10;

-- Natural key approach (avoid in warehouses): slower string join
-- JOIN dim_customer dc ON dc.email = fs.customer_email

세부 수준: 팩트 테이블의 상세 수준

-- Grain: one row per product per order line
-- Each row = one line item sold in one transaction
SELECT
  sale_id,
  date_key,
  product_key,
  quantity,
  unit_price,
  total_amount
FROM fact_sales
WHERE date_key = 20240315
ORDER BY sale_id;

가산적, 반가산적, 비가산적 측정값

SELECT
  dd.month_name,
  SUM(fs.total_amount)         AS total_revenue,   -- additive
  AVG(fs.unit_price)           AS avg_unit_price,   -- non-additive: use AVG
  SUM(fs.quantity)             AS total_units       -- additive
FROM fact_sales fs
JOIN dim_date dd ON dd.date_key = fs.date_key
WHERE dd.year = 2024
GROUP BY dd.month_name, dd.month_num
ORDER BY dd.month_num;

느리게 변하는 차원(SCD 유형 1 및 2)

-- SCD Type 2: add a new version of the row
ALTER TABLE dim_customer ADD COLUMN valid_from DATE;
ALTER TABLE dim_customer ADD COLUMN valid_to   DATE;
ALTER TABLE dim_customer ADD COLUMN is_current BOOLEAN DEFAULT TRUE;

-- Expire the old row
UPDATE dim_customer
SET is_current = FALSE,
    valid_to   = CURRENT_DATE - INTERVAL '1 day'
WHERE email = 'anna@example.com' AND is_current = TRUE;

-- Insert the updated version
INSERT INTO dim_customer (full_name, email, country, segment, valid_from, valid_to, is_current)
VALUES ('Anna Muller', 'anna@example.com', 'Austria', 'Premium', CURRENT_DATE, '9999-12-31', TRUE);

퇴화 차원

-- order_number is a degenerate dimension:
-- it lives in the fact table, no dim_order table needed
CREATE TABLE fact_order_lines (
  line_id      SERIAL PRIMARY KEY,
  order_number VARCHAR(20) NOT NULL,  -- degenerate dimension
  date_key     INT NOT NULL,
  product_key  INT NOT NULL,
  customer_key INT NOT NULL,
  quantity     INT NOT NULL,
  line_total   NUMERIC(12, 2) NOT NULL
);

SELECT order_number, SUM(line_total) AS order_total
FROM fact_order_lines
GROUP BY order_number
ORDER BY order_total DESC
LIMIT 5;

전체 스타 스키마 쿼리

SELECT
  dd.year,
  dd.quarter,
  dp.category,
  dc.country,
  SUM(fs.quantity)     AS units_sold,
  SUM(fs.total_amount) AS revenue
FROM fact_sales fs
JOIN dim_date     dd ON dd.date_key     = fs.date_key
JOIN dim_product  dp ON dp.product_key  = fs.product_key
JOIN dim_customer dc ON dc.customer_key = fs.customer_key
WHERE dd.year IN (2023, 2024)
  AND dp.category = 'Electronics'
GROUP BY dd.year, dd.quarter, dp.category, dc.country
ORDER BY dd.year, dd.quarter, revenue DESC;

빠른 확인: 팩트와 차원 비교

수업 요약