Why window functions are needed in SQL - Visual Breakdown

Concept Flow - Why window functions are needed

Start with a table of data

↓

Need to calculate aggregates per row

↓

Without window functions: use GROUP BY

↓

Problem: GROUP BY loses row details

↓

Result: detailed rows + aggregate info

Window functions let us calculate totals or ranks for each row without losing the original row details, unlike GROUP BY which groups rows and hides individual data.

Execution Sample

SQL

SELECT employee, department, salary,
       AVG(salary) OVER (PARTITION BY department) AS avg_dept_salary
FROM employees;

This query shows each employee's salary and the average salary of their department, keeping all employee rows.

Execution Table

Step	Action	Input Data	Window Function Calculation	Output Row
1	Read first employee row	{employee: 'Alice', department: 'Sales', salary: 5000}	Calculate AVG(salary) for Sales: (5000+6000)/2=5500	{Alice, Sales, 5000, 5500}
2	Read second employee row	{employee: 'Bob', department: 'Sales', salary: 6000}	Calculate AVG(salary) for Sales: 5500	{Bob, Sales, 6000, 5500}
3	Read third employee row	{employee: 'Charlie', department: 'HR', salary: 4000}	Calculate AVG(salary) for HR: 4000	{Charlie, HR, 4000, 4000}
4	All rows processed	N/A	N/A	Query complete with all rows and averages

💡 All rows processed, window function calculated average salary per department without grouping rows

Variable Tracker

Variable	Start	After 1	After 2	After 3	Final
Current Row	None	{Alice, Sales, 5000}	{Bob, Sales, 6000}	{Charlie, HR, 4000}	All rows processed
AVG(salary) OVER (PARTITION BY department)	None	5500 (Sales)	5500 (Sales)	4000 (HR)	Computed for each row

Key Moments - 2 Insights

Why can't we just use GROUP BY to get average salary per department?

GROUP BY combines rows into one per group, so you lose individual employee details. Window functions keep each row and add the aggregate info.

How does the window function know which rows to include in the average?

The PARTITION BY clause tells it to calculate the average separately for each department, so it only uses rows from the same department.

Visual Quiz - 3 Questions

Look at the execution table, what is the average salary for Sales at step 2?

A 5000

B 5500

C 6000

D 4000

💡 Hint

Check the 'Window Function Calculation' column at step 2 in the execution_table

At which step does the window function calculate the average salary for HR?

A Step 3

B Step 1

C Step 2

D Step 4

💡 Hint

Look for the row with department 'HR' in the 'Input Data' column in execution_table

If we removed PARTITION BY, what would happen to the average salary calculation?

A Average would be zero

B Query would fail with syntax error

C Average would be calculated over all employees, not per department

D Each row would show NULL for average

💡 Hint

PARTITION BY controls grouping for window functions, see concept_flow for grouping effect

Concept Snapshot

Window functions calculate aggregates like averages or ranks for each row
without grouping rows together.
Use OVER() with optional PARTITION BY to define groups.
This keeps all original rows and adds aggregate info.
Unlike GROUP BY, no rows are lost or combined.

Full Transcript

Window functions are needed because sometimes we want to calculate totals or averages for groups of rows but still keep each individual row visible. Normally, GROUP BY groups rows and returns one row per group, hiding individual details. Window functions solve this by calculating aggregates over a 'window' of rows related to each row, defined by PARTITION BY. This way, each row keeps its data and also shows the aggregate value for its group. For example, we can show each employee's salary and the average salary of their department in the same result. The execution steps show reading each row, calculating the average salary for that row's department, and outputting the row with the added average. This helps us understand why window functions are powerful and needed in SQL queries.

Practice

1. Why are window functions needed in SQL?
SELECT with window functions can do what that regular aggregation cannot?

easy

A. They speed up query execution by skipping indexes.

B. They allow calculations across rows without hiding individual rows.

C. They automatically create new tables for results.

D. They replace the need for JOIN operations.

Solution

1. Step 1: Understand aggregation vs window functions

   Regular aggregation groups rows and returns fewer rows, hiding details.

2. Step 2: Identify window function behavior

   Window functions calculate values across rows but keep all original rows visible.

3. Final Answer:

   They allow calculations across rows without hiding individual rows. -> Option B

4. Quick Check:

   Window functions keep rows visible = A [OK]

Hint: Window functions keep all rows, aggregations reduce rows [OK]

Common Mistakes:

• Thinking window functions create new tables
• Confusing window functions with JOINs
• Believing window functions speed up queries

2. Which of the following is the correct syntax to use a window function in SQL?

easy

A. SELECT name, RANK() OVER (ORDER BY score) FROM players;

B. SELECT name, RANK(score) FROM players ORDER BY score;

C. SELECT name, RANK() FROM players GROUP BY score;

D. SELECT name, RANK OVER (ORDER BY score) FROM players;

Solution

1. Step 1: Recall window function syntax

   Window functions use function() OVER (PARTITION BY ... ORDER BY ...).

2. Step 2: Check each option

   SELECT name, RANK() OVER (ORDER BY score) FROM players; correctly uses RANK() OVER (ORDER BY score). Others miss parentheses or OVER keyword.

3. Final Answer:

   SELECT name, RANK() OVER (ORDER BY score) FROM players; -> Option A

4. Quick Check:

   Window function syntax = function() OVER (...) [OK]

Hint: Window functions always use OVER() clause [OK]

Common Mistakes:

• Omitting parentheses after function name
• Forgetting the OVER keyword
• Using GROUP BY instead of window functions

3. Given the table sales with columns id, region, and amount, what will this query return?

SELECT id, region, amount, SUM(amount) OVER (PARTITION BY region ORDER BY id) AS running_total FROM sales ORDER BY id;

medium

A. A running total of sales amount per region, ordered by id, showing all rows.

B. Total sales amount per region, one row per region only.

C. Sum of all sales amounts ignoring region, one total value.

D. Syntax error due to missing GROUP BY clause.

Solution

1. Step 1: Understand SUM() OVER with PARTITION and ORDER

   SUM() OVER (PARTITION BY region ORDER BY id) calculates a running total per region ordered by id.

2. Step 2: Check output rows

   Window functions keep all rows, so all sales rows appear with running totals.

3. Final Answer:

   A running total of sales amount per region, ordered by id, showing all rows. -> Option A

4. Quick Check:

   Running total per group = A [OK]

Hint: SUM() OVER with PARTITION BY gives running totals per group [OK]

Common Mistakes:

• Thinking it returns one row per group
• Confusing window function with GROUP BY aggregation
• Expecting syntax error without GROUP BY

4. Identify the error in this SQL query using a window function:

SELECT employee_id, department, ROW_NUMBER() ORDER BY salary FROM employees;

medium

A. ORDER BY cannot be used inside ROW_NUMBER().

B. employee_id must be inside ROW_NUMBER() function.

C. ROW_NUMBER() requires GROUP BY clause.

D. Missing OVER() clause after ROW_NUMBER() function.

Solution

1. Step 1: Check window function syntax

   ROW_NUMBER() must be followed by OVER() with optional ORDER BY inside it.

2. Step 2: Identify missing OVER()

   The query uses ROW_NUMBER() ORDER BY salary without OVER(), causing syntax error.

3. Final Answer:

   Missing OVER() clause after ROW_NUMBER() function. -> Option D

4. Quick Check:

   Window functions need OVER() [OK]

Hint: Always add OVER() after window functions [OK]

Common Mistakes:

• Placing ORDER BY outside OVER()
• Thinking GROUP BY is needed for ROW_NUMBER()
• Misunderstanding function arguments

5. You want to find each employee's salary rank within their department, but also show all employees with their salaries and ranks. Which SQL approach best achieves this?

hard

A. Use a subquery to calculate ranks and join back to employees table.

B. Use GROUP BY department, salary with RANK() to get ranks per department.

C. Use RANK() OVER (PARTITION BY department ORDER BY salary DESC) to rank salaries per department while keeping all rows.

D. Use ORDER BY salary DESC with LIMIT to get top salaries per department.

Solution

1. Step 1: Understand the requirement

   We want ranks per department and all employees listed with their ranks.

2. Step 2: Evaluate options

   Use RANK() OVER (PARTITION BY department ORDER BY salary DESC) to rank salaries per department while keeping all rows. uses RANK() OVER (PARTITION BY department ORDER BY salary DESC) which ranks salaries per department and keeps all rows visible.

3. Final Answer:

   Use RANK() OVER (PARTITION BY department ORDER BY salary DESC) to rank salaries per department while keeping all rows. -> Option C

4. Quick Check:

   Window function with PARTITION keeps all rows and ranks [OK]

Hint: Use RANK() OVER PARTITION BY to rank within groups [OK]

Common Mistakes:

• Using GROUP BY loses individual rows
• Using subqueries unnecessarily complicates query
• LIMIT filters rows, not ranks all