AI and Machine Learning

Machine learning lets computers find patterns in data and turn them into predictions and decisions. Over four days you frame a real decision as a prediction problem, prepare the data, fit models, judge them, and ask whether they should be used at all. Each day carries its lecture slides, its dataset and its notebook. Day 1 is a mathematics workshop, and Day 2 runs as two lectures: machine learning foundations, then regression and classification. The materials for each day are shown below and are released as the course runs.

InstructorDr Fatih Kansoy

LengthFour teaching days

TeachingMonday to Thursday

FormatLectures, seminars and hands-on labs

AssessmentFriday, end of course

PrerequisitesNo programming assumed

Course materials

Slides The lecture deck for each day, released as the course reaches it. Datasets A page for each dataset, with every column and when it is known. Notebooks The executed notebook for each session, with every output kept in place. Syllabus Aims, learning outcomes, the four sessions and the reading list.

The data and the notebooks are available now. Every session's notebook and datasets, and a page documenting each dataset column by column, are on the datasets and code page. The lecture slides are being revised for the next cohort and are released as the course runs.

Course overview

The course follows one workflow from start to finish. You frame the decision, prepare the data, fit a model, and measure it against a baseline that costs nothing. Supervised methods come first, from linear and logistic regression through to trees, random forests and boosting. Unsupervised methods follow, where there are no labels to learn from.

Careful evaluation runs through all four days. A model can look strong because it has seen the answer already, so you learn to spot data leakage, to pick the metric that matches the decision, and to say what a prediction does not tell you about cause and effect.

Pythonscikit-learnJupyterUCI bank marketingNo programming assumed

Learning outcomes

By the end of the course you can:

frame a real decision as a supervised learning problem, and identify which variables are genuinely available before the decision is made;
explain and apply the core supervised models (linear and logistic regression, decision trees, random forests, and boosting) and the bias–variance trade-off that governs them;
evaluate models with metrics matched to the decision (accuracy, precision and recall, ROC–AUC, calibration) using train/validation/test splits and cross-validation;
diagnose data leakage, and distinguish predictive association from causal effect;
use unsupervised methods (K-means clustering and principal component analysis) to find and interpret structure in unlabelled data;
explain how embeddings, retrieval-augmented generation, and other modern AI workflows relate to classical machine learning;
assess the fairness, interpretability, monitoring and governance of a deployed model, and use AI tools responsibly, with evidence and reproducibility.

The four days

Day	Topic	Focus
Monday	Foundations: data, prediction, and trust	What machine learning is and is not; features and targets; loss; the train/test split; metrics; leakage; and why prediction is not causation.
Tuesday	Regression and classification: from models to decisions	Linear and logistic regression; coefficients and uncertainty; turning predicted probabilities into actions with cost-based thresholds.
Wednesday	Flexible models and evaluation	Decision trees, random forests, and boosting; cross-validation; precision, recall, ROC and PR curves; and auditing a model for data leakage.
Thursday	Unsupervised learning, modern AI, and responsible use	K-means clustering and PCA; embeddings and retrieval; and fairness, interpretability, monitoring, and governance.
Friday	Assessment	Assessment takes place on Friday.

Full syllabus, learning outcomes and reading list →

Oxford · United Kingdom Teaching CV