Data Science Practice

An original, hands-on portfolio of the core data-science skill set — Python, data wrangling, SQL, analysis, visualisation, and machine learning — built as nine self-contained mini-projects on public datasets.

This is an independent reimplementation of a widely taught data-science syllabus, written from scratch with new datasets, new code, and original explanations. It is not affiliated with, sponsored by, or endorsed by IBM or any other organisation; the certificate is credited only as the inspiration for the skill set. See NOTICE.md.

Start here

1. Clone the repository and install the environment:

   make install

2. Fetch the datasets (cached under data/raw/, which is git-ignored):

   python scripts/download_data.py --all

3. Verify everything with the offline test suite:

   make test

4. Open a module and work through its README, concepts.md, then its notebooks:

   jupyter lab 04-python-for-data-science/notebooks/

If you are new to the material, follow docs/learning-path.md. If you only want the map, the table below links every project.

Skill → project map

#	Skill area	Project	Dataset	Core techniques
01	What is data science	01-what-is-data-science	—	Lifecycle, roles, an applied case study
02	Tools for data science	02-tools-for-data-science	—	Notebooks, Markdown, Git, reproducibility
03	Data science methodology	03-data-science-methodology	Palmer Penguins	CRISP-DM end to end
04	Python for data science	04-python-for-data-science	Palmer Penguins	Types, functions, classes, files, NumPy, pandas, APIs
05	Python project	05-python-project	books.toscrape.com	Web scraping, cleaning, dashboard
06	Databases and SQL	06-databases-and-sql	OpenFlights	Normalised schema, joins, views, transactions, windows
07	Data analysis	07-data-analysis	California Housing	EDA, feature engineering, regression, evaluation
08	Data visualisation	08-data-visualization	Gapminder, USGS quakes	Matplotlib/Seaborn, Plotly/Dash, Folium
09	Machine learning	09-machine-learning	UCI Adult/Wine/Wholesale	Regression, classification, clustering, pipelines
Capstone	Applied capstone	falcon9-landing-analysis	SpaceX launch records	Full workflow (maintained separately)

Each module ships a README.md (objectives, prerequisites, dataset, topics, exercises, limitations), a plain-language concepts.md, worked solutions.md, and tutorial notebooks.

Learning path

flowchart LR
    A["01 Concepts"] --> B["02 Tooling"] --> C["03 Methodology"]
    C --> D["04 Python"]
    D --> E["05 Python project"]
    D --> F["06 Databases & SQL"]
    D --> G["07 Data analysis"]
    G --> H["09 Machine learning"]
    F --> H
    G --> I["08 Visualisation"]
    H --> J["Capstone"]
    I --> J

Loading

Repository structure

flowchart TD
    Hub["README.md<br/>skill-to-project map"]
    Hub --> Modules
    Hub --> Docs["docs/<br/>datasets, dictionary, glossary, structure"]
    Hub --> Shared
    subgraph Modules["Skill modules (01–09)"]
        direction LR
        Notebooks["notebooks/"]
        Teaching["README + concepts + solutions"]
    end
    subgraph Shared["Shared layer"]
        Pkg["src/ds_practice"]
        Scripts["scripts/"]
        Tests["tests/"]
    end
    Modules --> Pkg
    Shared --> Data["data/raw (git-ignored)"]

Loading

data-science-practice/
├── .github/workflows/ci.yml
├── 01-what-is-data-science/
├── 02-tools-for-data-science/
├── 03-data-science-methodology/
├── 04-python-for-data-science/
├── 05-python-project/
├── 06-databases-and-sql/
├── 07-data-analysis/
├── 08-data-visualization/
├── 09-machine-learning/
├── data/raw/                 # downloaded datasets (git-ignored)
├── docs/
├── scripts/
├── src/ds_practice/
├── tests/
├── CITATION.cff  LICENSE  Makefile  NOTICE.md
├── pyproject.toml  requirements.txt

Documentation map

Document	What it covers
docs/README.md	Documentation index and conventions
docs/course-map.md	Skill area to original project map
docs/learning-path.md	Suggested sequence, effort, and outcomes
docs/datasets.md	Source, licence, citation, and fetch command per dataset
docs/data-dictionary.md	Fields, types, units, and provenance
docs/methodology.md	How projects were built and kept reproducible
docs/glossary.md	Key terms defined
docs/STRUCTURE.md	Directory tree and architecture hierarchy
docs/FILE_CATALOG.md	Every tracked file with type and purpose
docs/erd-module-06.md	Entity-relationship diagram for module 06
NOTICE.md	Inspiration credit, non-affiliation, dataset attributions

Quickstart

make install     # create .venv and install pinned requirements
make data        # download every dataset into data/raw/
make test        # run the offline-safe test suite
make nb          # execute the offline-safe notebooks
jupyter lab      # explore the notebooks interactively

src/ds_practice provides the shared loaders, metrics, plotting theme, and SQL helpers, so no notebook copies data-loading or plotting code. Optional libraries (hdbscan, umap-learn) are used when present and skipped gracefully when not.

What this is

• A teaching portfolio: every module explains the idea, works an example on real data, and sets exercises with worked solutions.
• Reproducible: pinned dependencies, fixed random seeds, datasets fetched by one script, and a test for every module.
• Original: own code and prose on public datasets; attribution and non-affiliation are stated.

What this is not

• Not a copy of any course material, and not affiliated with the certificate that inspired it.
• Not a production system: models are illustrative and not tuned or monitored for deployment.
• Not a complete curriculum: it covers the core skill set and links to further reading rather than every advanced topic.

Capstone

The applied end-to-end project lives in a separate repository: falcon9-landing-analysis. It combines data acquisition, SQL, analysis, modelling, and visualisation into one workflow and is not duplicated here.

Licence and attribution

Code and prose are released under the MIT License. Datasets keep their own licences, recorded in docs/datasets.md and NOTICE.md.