dbt Training: From Installation to First Model

From installation to running your first model — with copy-pasteable commands throughout.

1. Introduction

dbt (Transforms landing data through staging → intermediate → marts.)dbt — Transforms landing data through staging → intermediate → marts. (data build tool) lets analysts and engineers transform data already loaded into a warehouse, using plain SQL SELECT statements. dbt handles the boilerplate — turning SQL into tables or views, managing dependencies, testing, and documentation.

2. Prerequisites

• Python 3.8+ installed
• A code editor (VS Code recommended)
• Access to a warehouse — Snowflake, BigQuery, Postgres, Redshift, or DuckDB (best for live demos, no cloud account needed)
• Git installed (optional, recommended for version control)

# check your python version
python --version

3. Installing dbt

dbt ships in two parts: dbt-core (the framework) and a dbt-<adapter> package (a connector for your specific warehouse).

3.1 Create a virtual environment

python -m venv dbt-env
source dbt-env/bin/activate      # Mac/Linux
dbt-env\Scripts\activate         # Windows

3.2 Install dbt-core plus an adapter

# Postgres
pip install dbt-core dbt-postgres
 
# Snowflake
pip install dbt-core dbt-snowflake
 
# DuckDB — best for local, zero-setup demos
pip install dbt-core dbt-duckdb

3.3 Verify the installation

dbt --version

You should see the dbt-core version and your installed adapter listed.

Core:
  - installed: 1.8.0
  - latest:    1.8.0 - Up to date!

Plugins:
  - postgres: 1.8.0 - Up to date!

4. Setting Up Your First Project

4.1 Initialize a new project

dbt init my_dbt_project

This prompts you to name the project, choose your adapter, and enter connection credentials — stored in profiles.yml (usually at ~/.dbt/profiles.yml).

4.2 Move into the project

cd my_dbt_project

4.3 Test the connection

dbt debug

Connection:
  ...
  Connection test: [OK connection ok]

All checks passed!

5. Project Structure

• my_dbt_project
  • dbt_project.yml
  • models
  • seeds
  • snapshots
  • macros
  • tests
  • analyses
  • profiles.yml

6. Creating Your First Model

A model is simply a .sql file with a SELECT statement. dbt compiles and runs it against your warehouse.

6.1 A simple model

models/customers.sql

-- models/customers.sql
select
    id as customer_id,
    first_name,
    last_name,
    email,
    created_at
from raw.customers
where email is not null

6.2 Load sample data as a seed (optional)

Add a CSV to seeds/ (e.g. seeds/raw_customers.csv), then:

dbt seed

6.3 Reference the seed from a model

-- models/stg_customers.sql
select
    id as customer_id,
    first_name,
    last_name
from {{ ref('raw_customers') }}

7. Running Your Model

# run all models
dbt run
 
# run one specific model
dbt run --select stg_customers
 
# run a model + everything downstream of it
dbt run --select stg_customers+
 
# run a model + everything upstream of it
dbt run --select +stg_customers
 
# full refresh (rebuild incremental models from scratch)
dbt run --full-refresh

1 of 1 START sql view model public.stg_customers .................. [RUN]
1 of 1 OK created sql view model public.stg_customers .............. [OK in 0.15s]

Completed successfully

Done. PASS=1 WARN=0 ERROR=0 SKIP=0 TOTAL=1

8. Testing Your Models

Built-in generic tests: unique, not_null, accepted_values, relationships.

8.1 Define tests

models/schema.yml

version: 2
 
models:
  - name: stg_customers
    columns:
      - name: customer_id
        tests:
          - unique
          - not_null

8.2 Run the tests

dbt test
 
# or scoped to one model
dbt test --select stg_customers

1 of 2 START test not_null_stg_customers_customer_id .......... [RUN]
1 of 2 PASS not_null_stg_customers_customer_id ................. [PASS in 0.10s]
2 of 2 START test unique_stg_customers_customer_id ............. [RUN]
2 of 2 PASS unique_stg_customers_customer_id ................... [PASS in 0.09s]

Completed successfully

Done. PASS=2 WARN=0 ERROR=0 SKIP=0 TOTAL=2

9. Documentation

dbt auto-generates a documentation site from your models and YAML descriptions.

# build the docs
dbt docs generate
 
# serve them locally in your browser
dbt docs serve

10. Command Cheat Sheet

11. Live Demo Flow

1. pip install dbt-core dbt-duckdb — install
2. dbt init demo_project — initialize
3. dbt debug — confirm the connection
4. Create models/customers.sql — write the first model
5. dbt run — run it
6. Add a test in schema.yml
7. dbt test — run it
8. dbt docs generate && dbt docs serve — show the documentation
9. Modify the model and rerun dbt run --select customers — show iteration speed

12. Common Pitfalls

Next

Continue with Lab 3: Medallion dbt Modeling — DHIS2 Case, which applies these fundamentals to the warehouse — modeling DHIS2 aggregate and tracker data through Bronze, Silver, and Gold layers.