Centre for Mathematical
Sciences
Wilberforce Road, Cambridge, CB3 0WB
Tel: (01223) 337958
Fax: (01223) 337956
Email: secretary [at] statslab.cam.ac.uk
All interested are welcome
This list is also available at talks.cam.
This list is subject to revision
Select a date to view the relevant seminar abstracts:
The Healthcare Commission is the independent regulator of the NHS in England, and carries out inspections, investigations, and annual ratings of all hospitals and primary care trusts. They have a commitment to the 'intelligent use' of the vast quantity of information to which they have access, and we have recently been collaborating in the development and application of appropriate statistical methods. Solutions to a range of challenges will be described: allowance for chance variability around targets set for MRSA reductions in hospitals, a major re-design of the entire hospital inspection regime into a 'risk-based' system, and the design of an automatic surveillance system for large numbers of indicators on hundreds of organisations. Technical issues include allowance for over-dispersion, hierarchical aggregation of performance indicators, risk-adjusted sequental analysis, and the control of false discovery rates.
We discuss a simple class of financial market models based on inhomogeneous telegraph processes. This model capture bullish and bearish trends as well as oversold/overbought market situations. The model under consideration is arbitrage-free if directions of jumps in stock prices are in a certain correspondence with their current velocity and interest rate behaviour. In the simplest case the model is complete. Diffusion rescaling of this model gives a natural representation of volatility. We provide explicit formulae for prices of standard European options are obtained, which permits to calculate directly implied volatilities with respect to various moneyness and maturity times of the option.
This model has a Markov-modulated implied volatility surface (see A.Jobert, L.C.G. Rogers, Option pricing with Markov-modulated dynamics). It gives an example of the implied volatility surface, which does not move by parallel shifts and which is based ona process different from exponential Levy (see L.C.G. Rogers, M.R.Tehranchi, The implied volatility surface does not move by parallel shifts.)
``Quantum phenomena are more truly random than any other phenomena one can imagine" (R. Gill, 1997). Do they follow Kolmogorov probability?
This talk has two parts. After a very brief review of Kolmogorov probability, we summarize Feynman's 1951 talk to the 2nd Berkeley symposium on mathematical statistics and probability in which he discussed applying Young's two-slit experiment to electrons. While his main point was to develop a new calculus of probabilities for the quantum world, he claimed that the two-slit experment gave empirical refutation to Kolmogorov's axiom of the additivity of probability of disjoint events. We discuss some of the more cogent and interesting critiques of this work.
In the second part, we describe basic aspects of the mathematical field of quantum probability in which events are subspaces of a Hilbert space rather than subsets of a sigma-field and random variables are observables, represented by Hermitian operators rather than measurable functions. Exploring these distinctions gives rise to a bestiary of differences from Kolmogorov probability. Conditional quantum probability allows us to view the two-slit experiment from a different perspective.
Prequential is a portmanteau word for predictive sequential a broad statistical methodology founded on a view of data (like Mark Twain.s view of history) as .just one darned thing after another.. A variety of statistical methods might be applied to learn projectible regularities in a data-sequence so as to improve forecasting; prequential analysis assesses how well this has been done by contrasting one-step ahead forecasts with realised outcomes.Many fundamental statistical concepts, such as consistency or efficiency, as well as techniques such as model selection, can be fruitfully redefined within this framework. This talk will outline the basic theory and some of its properties and applications.
The role of a statistician within an international company is diverse, working with interesting people in different areas. It provides different challenges (both technical and personal). This talk aims to give you a flavour of what life as an industrial statistian is like -- from the early months to a full career.
The Government Statistical Service is the UK~Rs largest provider of statisticians, with over 1200 professionals in more than 30 departments and agencies. This talk aims to give a general overview of the work of a Fast Stream Assistant Statistician, including examples of recent projects undertaken, as well as progression beyond the Fast Stream. Being a Government Statistician is an enjoyable and rewarding career, with excellent opportunities to work on high profile policy, often engaging with experts from across Government, Academia and Industry.
It is now 30 Years since I completed my PhD Dissertation on the "Evolution of Stochastic Automata" at the Cambridge Statistical Laboratory under the creative and inspirational leadership of Professor Peter Whittle. The seminar reviews the ways in which the conceptually rich theme of Stochastic Self-Organisation took me from working on strategic modelling projects for British Telecom in Cambridge to environmental networking in the Russian Arctic Kola Peninsula, as well as the World.s 1st Supersonic Car!I'll give overviews of projects that I worked on including KolaNet (Quick Response to possible Nuclear Accidents), ThrustSSC (1st Supersonic Car), EIGER (Computer Telephony Integration & Multimedia Networking), and the British Telecom LRPM (Long Range Planning Model). The topic of self-organising systems is still being actively researched in the context of neural, genetic and biological research programmes. The seminar briefly explores the ways in which a fuller understanding of stochastic self-organisation may further contribute to the development of business, government, education and society during the coming 20 to 30 years!
If a problem in functional data analysis is low-dimensional then the methodology for its solution can often be reduced to relatively conventional techniques in multivariate analysis. Hence, there is intrinsic interest in assessing the finite-dimensionality of functional data. We show that this problem has several unique features. From some viewpoints the problem is trivial, in the sense that continuously-distributed functional data which are exactly finite-dimensional are immediately recognisable as such, if the sample size is sufficiently large. However, in practice, functional data are almost always observed with noise, for example resulting from rounding or experimental error. Then the problem is almost insolubly difficult. In such cases a part of the average noise variance is confounded with the true signal, and is not identifiable. However, it is possible to define the unconfounded part of the noise variance. This represents the best possible lower bound to all potential values of average noise variance, and is estimable in low-noise settings. Moreover, bootstrap methods can be used to describe the reliability of estimates of unconfounded noise variance, under the assumption that the signal is finite-dimensional. Motivated by these ideas, we suggest techniques for assessing the finiteness of dimensionality. In particular, we show how to construct a critical point $\hat{v}_q$ such that, if the distribution of our functional data has fewer than q - 1 degrees of freedom, then we should be prepared to assume that the average variance of the added noise is at least $\hat{v}_q$ If this level seems too high then we must conclude that the dimension is at least q - 1. We show that simpler, more conventional techniques, based on hypothesis testing, are generally not effective.