Tables, Rows, Columns, Schemas, and Relational Thinking
Read a database you have never seen: state the grain of every table, find the join keys, and spot the join that multiplies rows.
SQL for Data Analysis and AI
A small analytics database, checked SQL queries, row-count reconciliation, and an AI-ready data evidence memo.
SELECT, WHERE, ORDER BY, and LIMIT
Turn a vague request into a question with one right answer, and learn the evaluation order that decides what your filter can see.
Data Types, Missing Values, and Safe Filtering
Missing data changes averages without warning. Learn to predict what NULL drops, and to publish the denominator alongside every number.
Aggregation with GROUP BY and HAVING
Summarise without losing the truth: pick the right grain, keep the awkward bucket visible, and make the parts sum to the whole.
Joins, Keys, Row Counts, and Duplicate Traps
The most expensive mistake in SQL: a join that repeats each order once per line item and inflates revenue by 3.4 times.
Subqueries and Common Table Expressions
Build long queries in named stages you can check one at a time, instead of one block that is either right or wrong.
Window Functions for Ranking, Running Totals, and Cohorts
Window functions keep every row while computing across them: running totals, rankings, and shares that find their own denominator.
Date and Time Analysis with Period Comparisons
A missing day looks exactly like a slow day. Learn the date-range convention that is correct for every period, every time.
Text Fields, Categorisation, and Feature Preparation
One city stored three ways counts as three cities. Find the variants, write a normalisation rule, and state what it does not catch.
Query Validation: Reconciliation, Edge Cases, and Sanity Checks
Re-running a query is not verification. Build checks that can actually fail, then write the evidence note that goes with the number.
SQL plus Python and pandas for Analysis Workflows
Aggregate in SQL, shape in pandas, and assert that both routes agree before anyone reports two different averages.
AI Data Preparation: Retrieval Tables, Metadata, Permissions, and Audit Fields
Retrieval quality is a data problem first: one row per citable thing, content that stands alone, and access enforced in the query.
Capstone: Analysis-Ready Dataset and SQL Evidence Memo
The capstone: one analysis-ready dataset, eight questions, every figure reconciled, and a memo that answers its own objections.
First-release boundary
This is a text-first course shell. Leonardo/image creation, image QC, dependency-heavy runtime QA, external review, and learner pilot evidence remain separate later quality steps.