Classifying News Headlines
Text classification from bag-of-words baselines to transformer embeddings, with a careful error analysis.
Showcase
At the end of each course, students publish their final data science project here as a short blog post: the question they asked, the data they used, what they found, and the code behind it.
First steps into data science — cleaning, exploring and visualising real data with pandas and matplotlib.
3 projects →End-to-end data science pipelines — modelling, evaluation and reproducible code.
3 projects →Text classification from bag-of-words baselines to transformer embeddings, with a careful error analysis.
Unsupervised learning on audio features to discover listener profiles, with a reproducible packaged codebase.
A full ML pipeline — feature engineering, cross-validation and model interpretation — on real estate transaction data.
Comparing pollution levels between cities and seasons using open government data.
Cleaning and exploring a movie ratings dataset to look for links between genre, budget and audience scores.
How does bike-sharing usage change with weather and time of day? A first exploration with pandas and matplotlib.