FILTER clause for conditional aggregation in PostgreSQL - Mini Project: Build & Apply

Using FILTER Clause for Conditional Aggregation in PostgreSQL

📖 Scenario: You are working with a sales database for a small store. The store wants to analyze sales data to see how many items were sold in different categories and how many sales were made on weekends.

🎯 Goal: Build a SQL query that uses the FILTER clause to count sales conditionally based on product category and day of the week.

📋 What You'll Learn

Create a table called sales with columns id (integer), category (text), amount (integer), and sale_date (date).

Insert the exact rows provided into the sales table.

Write a query that counts total sales and counts sales only for the category 'Books' using the FILTER clause.

Add a conditional count for sales made on weekends using the FILTER clause.

💡 Why This Matters

🌍 Real World

Stores and businesses often want to analyze sales data by different conditions, such as product category or day of the week, to make better decisions.

💼 Career

Knowing how to use conditional aggregation with FILTER clauses is useful for data analysts and database developers to write efficient and clear SQL queries.

1

Create the sales table and insert data

Create a table called sales with columns id (integer), category (text), amount (integer), and sale_date (date). Then insert these exact rows into the sales table:
(1, 'Books', 10, '2024-06-01'), (2, 'Electronics', 200, '2024-06-02'), (3, 'Books', 15, '2024-06-03'), (4, 'Clothing', 50, '2024-06-04'), (5, 'Books', 20, '2024-06-08')

PostgreSQL

-- Create the sales table and insert the given rows
-- Your code here

Hint

Use CREATE TABLE to define the table and INSERT INTO to add rows.

2

Add a helper variable for weekend days

Add a helper expression in your query to identify weekend days. Create a boolean expression called is_weekend that is true when sale_date is Saturday or Sunday. Use EXTRACT(DOW FROM sale_date) where 0 = Sunday and 6 = Saturday.

PostgreSQL

CREATE TABLE sales (
    id INTEGER,
    category TEXT,
    amount INTEGER,
    sale_date DATE
);

INSERT INTO sales (id, category, amount, sale_date) VALUES
(1, 'Books', 10, '2024-06-01'),
(2, 'Electronics', 200, '2024-06-02'),
(3, 'Books', 15, '2024-06-03'),
(4, 'Clothing', 50, '2024-06-04'),
(5, 'Books', 20, '2024-06-08');

-- Add a helper expression called is_weekend
-- Your code here

Hint

Use EXTRACT(DOW FROM sale_date) and compare it to 0 or 6 to find weekends.

3

Write a query using FILTER to count total and 'Books' sales

Write a SQL query that counts total sales using COUNT(*) and counts only sales where category = 'Books' using COUNT(*) FILTER (WHERE category = 'Books'). Use the sales table.

PostgreSQL

CREATE TABLE sales (
    id INTEGER,
    category TEXT,
    amount INTEGER,
    sale_date DATE
);

INSERT INTO sales (id, category, amount, sale_date) VALUES
(1, 'Books', 10, '2024-06-01'),
(2, 'Electronics', 200, '2024-06-02'),
(3, 'Books', 15, '2024-06-03'),
(4, 'Clothing', 50, '2024-06-04'),
(5, 'Books', 20, '2024-06-08');

-- Write a query to count total sales and count sales where category = 'Books' using FILTER
-- Your code here

Hint

Use COUNT(*) FILTER (WHERE condition) to count only rows matching the condition.

4

Add a FILTER clause to count weekend sales

Extend the previous query to add a new column weekend_sales that counts sales where the sale was made on a weekend. Use COUNT(*) FILTER (WHERE EXTRACT(DOW FROM sale_date) = 0 OR EXTRACT(DOW FROM sale_date) = 6).

PostgreSQL

CREATE TABLE sales (
    id INTEGER,
    category TEXT,
    amount INTEGER,
    sale_date DATE
);

INSERT INTO sales (id, category, amount, sale_date) VALUES
(1, 'Books', 10, '2024-06-01'),
(2, 'Electronics', 200, '2024-06-02'),
(3, 'Books', 15, '2024-06-03'),
(4, 'Clothing', 50, '2024-06-04'),
(5, 'Books', 20, '2024-06-08');

-- Add a FILTER clause to count weekend sales
-- Your code here

Hint

Use the same FILTER syntax with the weekend condition to count weekend sales.

Practice

1. What does the FILTER clause do in a PostgreSQL aggregate function?

easy

A. It filters the entire table before aggregation starts.

B. It applies a condition to include only certain rows in the aggregation.

C. It changes the grouping columns in the query.

D. It sorts the results after aggregation.

Solution

1. Step 1: Understand the purpose of FILTER in aggregation

   The FILTER clause allows you to specify a condition inside an aggregate function to include only rows that meet that condition.

2. Step 2: Compare with other options

   FILTER does not filter the whole table before aggregation, nor does it change grouping or sort results. It only limits rows counted or summed inside the aggregate.

3. Final Answer:

   It applies a condition to include only certain rows in the aggregation. -> Option B

4. Quick Check:

   FILTER = condition inside aggregate [OK]

Hint: FILTER applies conditions inside aggregates, not on whole table [OK]

Common Mistakes:

• Thinking FILTER filters rows before aggregation
• Confusing FILTER with WHERE clause
• Assuming FILTER changes GROUP BY columns

2. Which of the following is the correct syntax to count only rows where status = 'active' using FILTER in PostgreSQL?

easy

A. SELECT COUNT(*) FILTER (WHERE status = 'active') FROM users;

B. SELECT COUNT(*) FROM users FILTER (WHERE status = 'active');

C. SELECT COUNT(*) WHERE FILTER status = 'active' FROM users;

D. SELECT COUNT(*) FILTER WHERE status = 'active' FROM users;

Solution

1. Step 1: Recall FILTER syntax in aggregate functions

   The correct syntax places FILTER with parentheses and a WHERE clause inside the aggregate function: COUNT(*) FILTER (WHERE condition).

2. Step 2: Check each option

   SELECT COUNT(*) FILTER (WHERE status = 'active') FROM users; matches the correct syntax. The other options misuse FILTER placement or syntax.

3. Final Answer:

   SELECT COUNT(*) FILTER (WHERE status = 'active') FROM users; -> Option A

4. Quick Check:

   FILTER syntax = FILTER (WHERE condition) [OK]

Hint: Use FILTER (WHERE condition) inside aggregate parentheses [OK]

Common Mistakes:

• Omitting parentheses after FILTER
• Placing FILTER outside aggregate function
• Using FILTER without WHERE keyword

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

SELECT region, SUM(amount) FILTER (WHERE amount > 100) AS big_sales FROM sales GROUP BY region;

medium

A. Sum of all sales amounts per region.

B. Count of sales with amount greater than 100 per region.

C. Sum of sales amounts less than or equal to 100 per region.

D. Sum of sales amounts greater than 100 per region.

Solution

1. Step 1: Understand the FILTER condition in SUM

   The FILTER clause limits the SUM to only rows where amount > 100.

2. Step 2: Analyze the GROUP BY effect

   The query groups rows by region, then sums only amounts above 100 in each group.

3. Final Answer:

   Sum of sales amounts greater than 100 per region. -> Option D

4. Quick Check:

   SUM with FILTER (WHERE amount > 100) = sum of big sales [OK]

Hint: FILTER limits rows inside SUM to amount > 100 [OK]

Common Mistakes:

• Thinking it sums all amounts, ignoring FILTER
• Confusing SUM with COUNT
• Ignoring GROUP BY effect

4. Identify the error in this query:

SELECT department, COUNT(*) FILTER (status = 'active') FROM employees GROUP BY department;

medium

A. Missing WHERE keyword inside FILTER clause.

B. GROUP BY should include status column.

C. COUNT(*) cannot be used with FILTER.

D. FILTER clause must be outside aggregate function.

Solution

1. Step 1: Check FILTER clause syntax

   FILTER requires a WHERE keyword before the condition: FILTER (WHERE condition).

2. Step 2: Verify other parts

   GROUP BY on department is correct. COUNT(*) works with FILTER. FILTER must be inside aggregate function.

3. Final Answer:

   Missing WHERE keyword inside FILTER clause. -> Option A

4. Quick Check:

   FILTER syntax needs WHERE [OK]

Hint: Always include WHERE inside FILTER parentheses [OK]

Common Mistakes:

• Omitting WHERE in FILTER
• Placing FILTER outside aggregate
• Adding unnecessary columns to GROUP BY

5. You want to find the total sales and the total sales above 500 for each store in one query. Which query correctly uses FILTER to achieve this?

hard

A. SELECT store, SUM(amount) FILTER (WHERE amount > 500) AS total_sales, SUM(amount) AS big_sales FROM sales GROUP BY store;

B. SELECT store, SUM(amount) AS total_sales, COUNT(amount) FILTER (WHERE amount > 500) AS big_sales FROM sales GROUP BY store;

C. SELECT store, SUM(amount) AS total_sales, SUM(amount) FILTER (WHERE amount > 500) AS big_sales FROM sales GROUP BY store;

D. SELECT store, COUNT(amount) AS total_sales, SUM(amount) FILTER (WHERE amount > 500) AS big_sales FROM sales GROUP BY store;

Solution

1. Step 1: Identify total sales calculation

   Total sales is the sum of all amounts per store, so SUM(amount) without filter is correct.

2. Step 2: Identify big sales calculation

   Big sales are sums of amounts greater than 500, so SUM(amount) FILTER (WHERE amount > 500) is correct.

3. Step 3: Check options

   SELECT store, SUM(amount) AS total_sales, SUM(amount) FILTER (WHERE amount > 500) AS big_sales FROM sales GROUP BY store; correctly assigns total_sales and big_sales. The other options mix sums and counts incorrectly or swap columns.

4. Final Answer:

   SELECT store, SUM(amount) AS total_sales, SUM(amount) FILTER (WHERE amount > 500) AS big_sales FROM sales GROUP BY store; -> Option C

5. Quick Check:

   SUM with and without FILTER for totals [OK]

Hint: Use SUM(amount) and SUM(amount) FILTER (WHERE ...) together [OK]

Common Mistakes:

• Swapping filtered and unfiltered sums
• Using COUNT instead of SUM for amounts
• Misplacing FILTER clause