Introduction to SQL for Data Analytics

Why SQL is essential

Most analytics work happens where the data lives:

• Relational databases (PostgreSQL, MySQL, SQLite)
• Data warehouses (BigQuery, Snowflake, Redshift)

SQL is used to:

• Retrieve data efficiently
• Aggregate metrics (DAU, revenue, churn)
• Join multiple tables (users + orders + events)

Core mental model

• Data is stored in tables.
• A query reads rows/columns and returns a result set.

Minimal query

SELECT *
FROM users
LIMIT 10;

Typical analytics tables

• users(user_id, created_at, country, plan)
• orders(order_id, user_id, order_ts, amount)
• events(user_id, event_ts, event_name, device)

What you’ll learn in this phase

• SELECT, WHERE, ORDER BY, LIMIT
• Aggregations: COUNT, SUM, AVG
• GROUP BY, HAVING
• Joins
• Window functions
• CTEs
• Using SQL from Python (pandas)

Where SQL fits in an analytics workflow

You rarely run one query in isolation. The usual shape of the work is: raw tables live in a database, a query filters and reshapes them, and the result lands somewhere you can chart, model, or share.

diagram Where SQL fits in analytics mermaid

  flowchart LR
  A["users table"] --> D["SQL query"]
  B["orders table"] --> D
  C["events table"] --> D
  D --> E["Result set (rows + columns)"]
  E --> F["pandas DataFrame / BI dashboard"]

Raw tables live in a database; a query filters and shapes them into a result you can chart or model.

The database never changes because you queried it — a SELECT just reads and returns rows. That’s what makes SQL safe to explore with: you can run the same query a hundred times while you’re figuring out the right filter, and nothing gets damaged.

SQLite: a database in a single file

You don’t need a server to practice SQL. SQLite (Python’s built-in sqlite3 module) stores an entire database in one file — or even in memory — which makes it perfect for learning and for the exercises on this page.

A tiny in-memory database

import sqlite3
 
conn = sqlite3.connect(":memory:")
conn.execute("CREATE TABLE users (user_id INTEGER, country TEXT, plan TEXT)")
conn.executemany(
    "INSERT INTO users VALUES (?, ?, ?)",
    [(1, "IN", "free"), (2, "IN", "pro"), (3, "US", "pro")],
)
conn.commit()
 
rows = conn.execute("SELECT * FROM users").fetchall()
print(rows)

text

[(1, 'IN', 'free'), (2, 'IN', 'pro'), (3, 'US', 'pro')]

Analyst’s tip

Every exercise in this SQL phase runs against a throwaway sqlite3.connect(":memory:") database — no setup file, no server, nothing left behind. That’s exactly how you should prototype real analytics queries too: build the query against a small sample first, confirm the shape of the result, then point it at the real warehouse table.

🧪 Try It Yourself

Exercise 1 – Create a table and read it back

pch.exerciseTag pch.exerciseTitle

Exercise 2 – Filter rows with WHERE

pch.exerciseTag pch.exerciseTitle

Exercise 3 – Load a query result into pandas

pch.exerciseTag pch.exerciseTitle

Next

Head to SQL Basics (SELECT, WHERE, ORDER BY, LIMIT) to practice the four clauses you’ll use in almost every query you ever write.