Roots Academy 2026 — Data Quality

Materials for the Data Quality morning session at Roots Academy 2026 (Dataroots).

Instructor: Eya Akrimi · Data Engineer @ Dataroots When: Wed 23 Sept 2026 · 9:00 → 12:00 Format: ~2h theory + ~50min hands-on with Soda Core

You will need Python 3.10+ and git. Everything else is installed by uv sync (or pip install -e .).

# 1. Clone
git clone https://github.com/datarootsio/rootsacademy-2026-data-quality.git
cd rootsacademy-2026-data-quality

# 2. Install (choose one)
uv sync                              # ← fast, recommended
# or
python -m venv .venv && source .venv/bin/activate && pip install -e .

# 3. Generate the dirty dataset
python data/generate_dirty_data.py

# 4. First scan (should show 4 fails, 1 pass — that's the point)
soda scan -d avocados -c soda/configuration.yml soda/checks_starter.yml

If the last command prints check results, you're ready — close the terminal until 11am.

.
├── README.md
├── pyproject.toml            ← deps: soda-core-duckdb, pandas, jupyter
├── data/
│   └── generate_dirty_data.py    ← creates avocados.duckdb with realistic dirt
├── soda/
│   ├── configuration.yml         ← the DuckDB data source
│   ├── checks_starter.yml        ← the 5 checks we run first
│   └── checks_extended.yml       ← more checks for exercise 2
├── exercises/
│   ├── 01_first_scan.md          ← guided walkthrough
│   ├── 02_add_your_checks.md     ← write your own
│   ├── 03_break_it_on_purpose.md ← inject bad data, watch it fail
│   └── notebook.ipynb            ← optional Jupyter path
├── solutions/
│   ├── checks_solution.yml
│   └── SOLUTIONS.md              ← DO NOT PEEK 👀
├── slides/
│   └── DataQuality-RA2026.pptx
└── .github/workflows/
    └── soda-ci.yml               ← DQ in CI (bonus discussion at the end)

Soda Core is an open-source data quality framework. You write human-readable checks in YAML (SodaCL) — things like "no missing values in region" or "row count > 0" — and soda scan runs them against a database. It supports DuckDB, PostgreSQL, Snowflake, BigQuery, Databricks and more. If you like what you see today, its bigger sibling Soda Cloud adds a UI, scheduling, notifications and a data catalog integration — same YAML checks, hosted.

Builds on the prior Dataroots RA data-quality sessions (2023 Great Expectations by the KBC team; 2024 Data Quality). Story arc & 6-dimensions taxonomy adapted from those decks; hands-on rebuilt around Soda + DuckDB so nothing needs cloud credentials.

MIT — see LICENSE.