A core part of the first year is the first-year project. The project is completed through Term 3 and the Summer. These projects often lead into the main research of the PhD thesis, but students also have the opportunity to amend and change the research topic if appropriate.

Students choose from a wide selection of projects, and also have the opportunity to develop their own project alongside academics, covering the whole spectrum of statistics, including but not limited to areas within:

• Applied (Interdisciplinary) Statistics

• Computational Statistics and Machine Learning

• Mathematical Finance

• Probability

• Statistical Methodology and Theory

Examples of projects offered

Applied and Interdisciplinary Probability and Statistics

• Characterising uncertainty in radiotherapy towards treatment optimisation
• Deterministic versus stochastic modelling in infectious disease epidemiology
• Statistical Innovations in the Social Sciences

Computational Statistics and Machine Learning

• Improving off-line deep reinforcement learning through uncertainty quantification

• Statistical Machine Learning, Robustness, Causality, Decentralised AI

• Feature allocation models in Bayesian nonparametric statistics and machine learning

• Advancing deep reinforcement learning methods to unlock real-world applications

Mathematical Finance

• (U, ρ)-portfolio selection and (U, ρ)-arbitrage
• Stochastic Control with Partial Information
• Risk measurement and management in finance and actuarial science
• Optimal Stopping in Behavioural Economics
• Applications of stochastic control theory in portfolio selection

Probability

• Interacting particle systems, metastability, and mixing times
• Reflected Brownian motion in unbounded domain
• Interacting particle systems, large deviations, metastability, and mixing times
• Stochastic processes with path discontinuities
• Rough path theory with jumps
• Balanced Allocation Meets Queueing Theory
• Spatial branching structures in compact domains and boundary selection
• Statistical inference, rough paths, multiscale systems

Statistical Methodology and Theory

• Bayesian Analysis of Latent Variable Network Models
• Tests of MCAR based on observed covariance matrices
• Local differential privacy in time series data
• Inference for SDEs based on advanced numerical schemes
• Predictive inference: Generalised copula trained using prequential scoring rules
• Change point localisation and inference
• The genealogy of a multilevel splitting algorithm