Ben Taylor

Contact Details

Benjamin Taylor, MSci MSc PhD
Senior Research Associate
Faculty of Health and Medicine
B27 Furness Building
Lancaster University
Lancaster
LA1 4YF
UK

Telephone: +44 (0)1524 593499

Email: b.taylor1@lancaster.ac.uk
Department Website: http://www.lancs.ac.uk/shm/med/research/chicas.php
Personal Website: (You're here!)

About me

I am currently a research associate working with the CHICAS group led by Professor Peter Diggle. Formerly, I was a PhD student under the supervision of Professor Paul Fearnhead. My background is in pure mathematics and medical statistics.

Research

I work in the area of computational statistics. I am currently working collaboratively on spatio-temporal statistical methods, associated computational algorithms and the integration of these into web-based information systems. Specifically, I'm interested in,

Methodological and computational aspects of log-Gaussian Cox Processes.
Forecasting meningitis incidence in sub-Saharan Africa in collaboration with the World Health Organisation.
Spatial prediction of campylobacter in the UK.

My PhD research concerned particle filtering (Sequential Monte Carlo) methodology - in particular: adaptive SMC, forward-backward algorithms, fixed parameter estimation and the use of SMC in fixed data scenarios. As part of my PhD, I also conducted research on the rating of NCAA college basketball teams and NBA players.

Publications

In Progress / Pre-Prints

Spatial and Spatio-Temporal Log-Gaussian Cox processes: Extending the Geostatistical Paradigm. Peter J Diggle, Paula Moraga, Barry Rowlingson, Benjamin M. Taylor. Submitted
Corrigendum: Spatiotemporal Prediction for Log Gaussian Cox Processes. Benjamin M. Taylor and Peter J. Diggle. To apprear in JRSSB.
Towards Real-time Spatio-temporal Prediction of District-level Meningitis Incidence in Sub-Saharan Africa. Michelle C. Stanton, Lydiane Agier, Benjamin M. Taylor, Peter J. Diggle. Submitted
INLA or MCMC? A Tutorial and Comparative Evaluation for Spatial Prediction in log-Gaussian Cox Processes. B. M. Taylor and P. J. Diggle. Submitted.
An Adaptive Sequential Monte Carlo Sampler P. Fearnhead and B. M. Taylor. To appear in Bayesian Analysis.

Journal Articles

2013

lgcp - An R package for Inference With Spatiotemporal Log-Gaussian Cox Processes. B. M. Taylor, T. M. Davies, B. S. Rowlingson, P. J. Diggle. Journal of Statistical Software Vol. 52 (2013) Iss. 4. See also 'Software' section below, or link to CRAN
Mapping English GP Prescribing Data: a tool for monitoring health-service inequalities. B. S. Rowlingson, E. Lawson, B. M. Taylor and P. J. Diggle. BMJ Open Vol. 3 (2013), Iss. 1.

2011

On Estimating the Ability of NBA Players P. Fearnhead and B. M. Taylor. Journal of Quantitative Analysis in Sports. Vol. 7 (2011), Iss. 3. Article 11.

2010

Calculating Strength of Schedule, and Choosing Teams for March Madness P. Fearnhead and B. M. Taylor. The American Statistician, Vol. 64 (2010), No. 2:108-115.

Comments and Contributions

Comment on paper by N. Chopin and P. Jacob: "Free Energy Sequential Monte Carlo, Application to Mixture Modelling". B. M. Taylor. Bayesian Statistics 9.

Theses

Sequential Methodology and Applications in Sports-Rating. PhD Thesis. Lancaster University (2010).
Simultaneous Modelling of Longitudinal and Time-to-Event Data. MSc Dissertation. University of Leiceseter (2006).
Representation Theory and Characters of Groups. MSci Dissertation. University of Birmingham (2004).

Software

lgcp - version 0.9-5 Package News

lgcp - An R package for Inference With Spatio-temporal Log-Gaussian Cox Processes. B. M. Taylor, T. M. Davies, B. S. Rowlingson, P. J. Diggle LINK TO CRAN LINK TO PACKAGE VIGNETTE
Improved computation time.
Prediction for spatial log-Gaussian Cox processes.

miscFuncs - version 1.2 Package News

Miscellaneous Useful Functions: LaTeX tables, Kalman filter, web scraping, development tools. LINK TO CRAN
Package vignette: Kalman filtering in miscFuncs

Presentations

As well as giving talks in the department here as part of the Computational Statistics Group and at statistics forums, I have presented my research to a wider audience:

Spatio-temporal Log-Gaussian Cox Processes - Properties, Inference and an R Package B. Taylor, P. Diggle, B. Rowlingson, T. Davies
Particle Independent Metropolis-Hastings From the PMCMC paper by Andrieu et al B. Taylor Computational Statistics Group Meeting. Feb 16th 2010. [LINK TO SLIDES]
An Adaptive Sequential Monte Carlo Algorithm For Bayesian Mixture Analysis B. Taylor, P. Fearnhead University of Cambridge. Nov 4th 2009
Adaptive Sequential Monte Carlo in Mixture Analysis B. Taylor, P. Fearnhead SAMSI Particle Learning Working Group. Jan 22nd 2009
Fixed Parameter Estimation in Static Sequential Monte Carlo Models With Application to Motor Unit Number Estimation. B. Taylor, P. Fearnhead, A. Pettitt. World Meeting of The International Society For Bayesian Analysis. Hamilton Island. July 2008 [LINK TO ABSTRACT]
Adaptive Sequential Monte Carlo Methods For Static Inference in Bayesian Mixture Analysis. B. Taylor, P. Fearnhead, A. Pettitt. University Of Warwick. June 2008

Posters

I presented three posters during Autumn term 2008/2009, two as part of Sci-Tech Graduate School initiatives and one at the RSS HQ in London. Other posters include the following:

Bayesian Parameter Estimation for the Log-Gaussian Cox Process. P. Diggle, B. Taylor, T. Davies, B. Rowlingson INFER 2011, Warwick University. March 2011
Dynamic Modelling for Wind Prediction. R. Griffiths, B.Taylor. STORi Summer Internship poster. Lancaster University. August 2010.
An Adaptive Sequential Monte Carlo Sampler. B. Taylor, P. Fearnhead Valencia/ISBA World Meeting. June 2010

Workshops and Conferences

Inference For Epidemic-related Risk (INFER). March 2011. Uiversity of Warwick.
Valencia 9 / ISBA World Meeting. June 2010. Benidorm.
Advanced Use of R. Lancaster University. April 2010.
Introduction to C++ and Advanced C++ Programming. Lancaster University. Feb 2010.
Master Class: Introduction to Modern Smoothing Methods: GAMLSS and P-Splines in Action. Lancaster University. December 2009.
One Day Workshop on Nonlinear Methods for Estimation. Lancatster University. May 2009.
EPSRC Workshop/Symposium on Markov Chain Monte Carlo. March 2009.
Lancaster University Faculty of Science and Technology Christmas Conference. December 2009.
ISBA World Meeting, Hamilton Island. July 2008.
Workshop on the Applied Mathematics/Statistics Interface. University Of Warwick. June 2008.
APTS 2007/2008.

Awards

ISBA 2008 Travel Award for Postgraduate Students. This was a competatively awarded prize based on abstract submission.

Other Activities in the Mathematics Department

As a PhD student, tutored on various undergraduate courses including first and second year probability and statistics, second and third year likelihood theory and stochastic processes. I have also tutored on as well as led the R component of MATH 390 (3rd year project). I was postgraduate rep in the 2008-2009 academic year.

I was heavily involved in the organisation of The Research Students' Conference in Probability and Statistics, which was held here in Lancaster in March 2009. I was responsible for finance and sponsorship and also acted as secretary at the meetings.

During the summer of 2010, I supervised a STORi summer internship project, Dynamic Modelling for Wind Prediction.

I designed and organised the R course for STORi summer interns in summer 2010; this involved writing learning materials and facilitating the teamwork exercises as well as a competitive team programming challenge.

In September 2010, I acted as a course tutor for the APTS Nonparametric Smoothing module.

Personal Interests

Music

Probably my main past-time. I play guitar - electric, steel-strung acoustic and classical to some discernable level of competency. I also play violoncello, electric bass, drums, keyboard, digeridoo; and to a lesser level of competence, mandolin and saxophone - a lot of space in our house is taken up by instruments!

Fitness and Healthy Living

I swim, run and cycle regularly.

Other stuff

When I get time, I enjoy other artistic pursuits such as woodwork, drawing and painting (some examples below).

You can do more than just statistics with R

I also like to code in R for fun, some examples are below.

Mandlebrot Set

This program plots the Mandlebrot set to any desired level of accuracy (memory permitting obviously) and allows the user to zoom in on areas of interest.

Sudoku Solver

Less hassle than doing them by hand ... only does the easy ones at the moment though.

Taylor Series

Here is some code to demonstrate a Taylor series approximation to the function y=sin(x). The function ts(n,a) returns a Taylor series approximation up to the nth derivative at the point a (which in this example should be between -5 and 5). As n is increased, the approximation gets better.

rm(list=ls())

x <- seq(-5,5,length.out=1000)

diff <- function(n,a){
   if(n%%4==1){
     return(cos(a))
   }
   else if(n%%4==2){
     return(-sin(a))
   }
   else if(n%%4==3){
     return(-cos(a))
   }
   else{
     return(sin(a))
   }
}

ts <- function(n,a){
   terms <- matrix(NA,n,length(x))
   for (i in 1:n){
     terms[i,] <- diff(i,a) * (1/factorial(i))*(x-a)^i
   }
   return(sin(a) + colSums(terms))
}

y <- sin(x)

plot(x,y,type="l")

tsapprox <- ts(2,-2)

lines(x,tsapprox,col="red",lty="dashed")