Analytics Engineering: Advanced SQL, Data Modeling & dbt Project

Start with PROJECT_OVERVIEW.md for what you're building and why. This file (README.md) is where the step-by-step setup lives.

Due: 6 days, run as a sprint, plus a required share-out session scheduled after. See CHECKLIST_TIMELINE.md for the day-by-day pace and the full submission checklist.

This repo is a GitHub template — a starting point, not something you edit directly on Marcy's copy of it.

Getting started

Step 1: Get your own copy

On this repo's GitHub page, click "Use this template" → "Create a new repository" (not Fork — Fork keeps a visible link back to this template, which isn't what you want for a portfolio project). Name it something like analytics-engineering-dbt, keep it public, and create it.

Step 2: Clone your new repo locally

git clone <the URL of your own new repo>
cd <your-repo-name>

Step 3: Confirm your environment — the basics are already set up

Like Module 3/6, .gitignore, LICENSE, and a real git history are already here — git itself isn't being newly tested this module:

ls -a          # should show .gitignore among the files
cat LICENSE    # should show the MIT License text
git log --oneline

One real edit still needed: open LICENSE and replace the placeholder [YOUR NAME] on the copyright line with your actual name. Commit that change alongside your other early commits.

Step 4: Set up dbt in a dedicated virtual environment

See GETTING_STARTED.md — dbt is new this module and gets its own dedicated setup walkthrough, including why a dedicated virtual environment matters here specifically, not just as generic advice.

Step 5: Reconnect to your Module 3 database

Also in GETTING_STARTED.md — the same database you built in Module 3 (and reconnected to again in Module 4), no new domain data this time.

What to do

• starter/dbt_project/ has the given project skeleton: dbt_project.yml (already configured), a models/staging/, models/intermediate/, models/marts/ folder structure (each with a short README explaining that layer's real job), and models/metricflow_time_spine.sql (a given, mechanical file dbt's semantic-layer feature needs — not graded content).
• Fill in starter/required_components.md as you build, not after — it's the real record of what you actually did (including the test you broke on purpose and the real EXPLAIN ANALYZE before/after), not a final summary written from memory.
• Build staging models for your own raw Module 3 tables — cleaning only.
• Build ≥1 intermediate model with the real business logic (joins, cross-table calculations) that doesn't belong in staging.
• Build ≥1 mart table answering a real business question from your own domain's Module 3 SCENARIOS.md entry.
• Write ≥3 real schema.yml tests; verify at least one genuinely fails on bad data, not just that it passes on good data.
• Define ≥1 semantic metric, centrally, in models/marts/ (or wherever your semantic model lives) — reuse it, don't recalculate it elsewhere.
• Run a real EXPLAIN ANALYZE on one of your own slow(ish) queries, apply a real fix, measure the real difference.
• Run dbt docs generate for real; fill in starter/data_dictionary.md; trace one real mart column back to its raw source via the lineage graph.

CHECKLIST_TIMELINE.md has the suggested day-by-day pace and the full sequenced checklist.

Where's the exact bar for "done," and what are the optional stretch goals? This repo (your own copy) doesn't include MVP.md (your Minimum Viable Product — the required baseline) or ABOVE_AND_BEYOND.md on purpose — they're not something to keep sitting in your portfolio repo. Ask your instructor for the link to this template's project-scope branch to read them, or check the checklist your instructor shares through the classroom, which covers the same ground.