Applied ML & Data Science
Hagnýtt vélanám og gagnavísindi
Real data, real models, real judgement.
Sign up on AblerWhat they'll learn
- Real, messy data end to end Take an actual dataset from raw mess to finished result, the whole pipeline.
- Features and model choice Build useful features and choose the right model for the problem.
- Honest evaluation Evaluate rigorously and document limits with proper model cards.
- Ethics and the EU AI Act Apply real AI ethics and the EU AI Act to the work, not just in theory.
A full data-science project on a real dataset, documented with a professional model card.
The 13-week journey
-
Sprint 1 — The data pipeline
Messy real data: cleaning, missing values, leakage; exploratory analysis and visualisation; framing a question a model can answer. Ends with a cleaned dataset and a clear question.
-
Sprint 2 — Modelling well
Feature engineering, model selection, cross-validation, and choosing metrics that fit the problem; comparing models honestly. Ends with a tuned, defensible model.
-
Sprint 3 — Responsible & done
Bias analysis, the EU AI Act and data-protection basics, and a professional model card; packaging the project for a portfolio. Ethics through real cases.
-
Week 13 — Showcase
Project presentations with results, limitations and a portfolio-ready write-up.
What we cover
Every topic, unit by unit — so you know exactly what your child builds and learns.
The data pipeline
- Cleaning messy data, missing values and leakage
- Exploratory analysis and visualisation
- Framing a question a model can answer
- A cleaned dataset and a clear question
Modelling well
- Feature engineering
- Model selection and cross-validation
- Choosing metrics that fit the problem
- Comparing models honestly
Responsible & done
- Bias analysis on real data
- EU AI Act and data-protection basics
- Writing a professional model card
- Packaging for a portfolio
Showcase & portfolio
- Presenting results and limitations
- A portfolio-ready write-up
- Project presentations to the audience
A real-world data project with its analysis, model and honest evaluation — presented like a junior data scientist.
The data-science track for teens who want depth: real, messy data taken from question to model to honest conclusion — with the evaluation rigour, ethics and documentation a professional brings. The course that turns AI enthusiasm into method.
The hooks
Teen hook: “I took a real dataset from mess to a model I can defend — and I know exactly what it can’t do.” Parent hook: “Genuine data-science skills and ethics — university-level substance, taught by building.”
Who it’s for
16–18s with Python and some ML (AI Studio or Machine Learning Lab + placement). Thrives: the analytical, the rigorous, the future researchers and ML practitioners.
Outcomes — by the end, students can
Clean and explore messy real data; build features and select models; evaluate with appropriate metrics and cross-validation; analyse bias; write a model card; situate work within the EU AI Act and data-protection basics.
Tools & compliance
Python notebooks, pandas/scikit-learn (and a deep-learning taster), real and open datasets, own accounts at 16+; ethics, data-protection and honest-evaluation norms enforced.
Where this course fits
A technical companion to AI Studio; strong preparation for a data-driven Launchpad capstone.
Parent questions
How is this different from AI Studio?
AI Studio spans ML, neural nets and LLM apps; this goes deeper on the data-science craft — messy data, feature work, evaluation and ethics.
Is it useful for university?
Very — a documented data project and honest evaluation are exactly what STEM and CS applications value.
What background is needed?
Python and some ML (AI Studio or Machine Learning Lab, or placement).