Add SQL examples for DENSE_RANK() usage

07_Window_Functions/03_DENSE_RANK.sql

@@ -0,0 +1,186 @@
+-- ==========================================================
+-- SQL Engineering Handbook
+--
+-- Topic:        DENSE_RANK()
+-- Module:       07_Window_Functions / 03_DENSE_RANK
+-- Difficulty:   Beginner
+-- Author:       SQL Engineering Handbook Contributors
+-- Description:  Assigns a competition rank with NO gaps after ties,
+--               unlike RANK(). [1, 2, 2, 3, 4, 5]
+-- Prerequisites: 01_ROW_NUMBER, 02_RANK
+-- Dataset:      employes
+-- Learning
+-- Objectives:   1. Assign gap-free ranks.
+--               2. Compare DENSE_RANK() against RANK().
+--               3. Filter dense-ranked results with a CTE.
+-- ==========================================================
+
+
+-- ==========================================================
+-- Q1 -- Assign Dense Rank to Employees Ordered by manager_id
+-- ==========================================================
+-- Problem Statement:
+--   Assign a dense rank to employees ordered by manager_id ascending.
+
+SELECT
+    emp_id,
+    emp_name,
+    DENSE_RANK() OVER (ORDER BY manager_id ASC) AS dense_ranks
+FROM employes;
+
+-- Business Use Case:
+--   Group employees into consecutive "manager tiers" for an org chart
+--   visualization, where no tier number should be skipped.
+
+
+-- ==========================================================
+-- Q2 -- Show Employee Name and Dense Rank
+-- ==========================================================
+-- Problem Statement:
+--   Return emp_name and its dense rank, ordered by emp_id ascending.
+
+SELECT
+    emp_name,
+    DENSE_RANK() OVER (ORDER BY emp_id ASC) AS dense_ranks
+FROM employes;
+
+
+-- ==========================================================
+-- Q3 -- Show Employees With Dense Rank <= 3
+-- ==========================================================
+-- Problem Statement:
+--   Return employees whose dense rank (by emp_id) is 3 or better.
+
+WITH dense_ranks AS (
+    SELECT
+        emp_id,
+        emp_name,
+        DENSE_RANK() OVER (ORDER BY emp_id) AS d_ranks
+    FROM employes
+)
+SELECT
+    emp_id,
+    emp_name,
+    d_ranks
+FROM dense_ranks
+WHERE d_ranks <= 3;
+
+-- Variant -- same question, ranked by manager_id descending instead,
+-- to show how the ORDER BY direction changes which rows qualify:
+
+WITH dense_ranks_by_manager AS (
+    SELECT
+        emp_id,
+        emp_name,
+        manager_id,
+        DENSE_RANK() OVER (ORDER BY manager_id DESC) AS d_ranks
+    FROM employes
+)
+SELECT
+    emp_id,
+    emp_name,
+    d_ranks
+FROM dense_ranks_by_manager
+WHERE d_ranks <= 3;
+
+-- Business Use Case:
+--   Retrieve the top 3 consecutive salary/seniority bands without
+--   skipping a band number, which is important for reports that must
+--   show every tier that exists in the top bracket.
+
+
+-- ==========================================================
+-- Q4 -- Compare RANK() and DENSE_RANK()
+-- ==========================================================
+-- Problem Statement:
+--   Show emp_id, emp_name, manager_id, RANK(), and DENSE_RANK() side
+--   by side to visually confirm the gap difference.
+
+WITH ranked_employees AS (
+    SELECT
+        emp_id,
+        emp_name,
+        manager_id,
+        DENSE_RANK() OVER (ORDER BY manager_id DESC) AS d_ranks,
+        RANK()       OVER (ORDER BY manager_id DESC) AS ranks
+    FROM employes
+)
+SELECT
+    emp_id,
+    emp_name,
+    ranks,
+    d_ranks
+FROM ranked_employees;
+
+-- Expected Output (conceptual):
+-- manager_id | ranks | d_ranks
+-- -----------|-------|--------
+--     9      |   1   |   1
+--     9      |   1   |   1     <- tie
+--     7      |   3   |   2     <- RANK jumps to 3, DENSE_RANK moves to 2
+--
+-- Explanation:
+--   RANK() "spends" rank numbers on tied rows; DENSE_RANK() does not.
+
+
+-- ==========================================================
+-- Q5 -- Show the Employee(s) With Dense Rank = 1
+-- ==========================================================
+-- Problem Statement:
+--   Return the employee(s) sitting at dense rank 1 (highest manager_id).
+
+WITH ranked_employees AS (
+    SELECT
+        emp_id,
+        emp_name,
+        manager_id,
+        DENSE_RANK() OVER (ORDER BY manager_id DESC) AS d_ranks,
+        RANK()       OVER (ORDER BY manager_id DESC) AS ranks
+    FROM employes
+)
+SELECT
+    emp_id,
+    emp_name,
+    ranks,
+    d_ranks
+FROM ranked_employees
+WHERE d_ranks = 1;
+
+-- Business Use Case:
+--   Identify all employees reporting to the highest-ID manager in a
+--   single, readable query -- reusable for any "top tier" lookup.
+
+
+-- ==========================================================
+-- Summary
+-- ==========================================================
+-- DENSE_RANK() gives consecutive ranks with no gaps, even after ties.
+-- Choose it whenever the business needs a compact tier numbering
+-- scheme (e.g., pay grades, pricing tiers, risk bands).
+
+-- ==========================================================
+-- Common Mistakes
+-- ==========================================================
+-- 1. Assuming DENSE_RANK() and RANK() are equivalent on tie-free data
+--    -- they only diverge once ties appear.
+-- 2. Using DENSE_RANK() when the business actually wants RANK()'s
+--    "gap = number of ties" semantics.
+
+-- ==========================================================
+-- Performance Notes
+-- ==========================================================
+-- Cost profile is identical to RANK() -- a single sorted pass with a
+-- per-row tie comparison. No additional overhead versus RANK().
+
+-- ==========================================================
+-- Interview Questions
+-- ==========================================================
+-- 1. Give a real-world example where DENSE_RANK() is clearly the
+--    correct choice over RANK().
+-- 2. How would you find the "second highest" distinct salary using
+--    DENSE_RANK()?
+
+-- ==========================================================
+-- Further Reading
+-- ==========================================================
+-- https://dev.mysql.com/doc/refman/8.0/en/window-function-descriptions.html#function_dense-rank