I am a freelance chartered statistician specialising in data visualisation and Bayesian modelling. I provide training courses, coaching and consultancy. I've helped analysts from stats and machine learning backgrounds, visualisation/design people, economists, health and social care professionals, academics, civil servants, journalists and more. I'm also interested in helping managers responsible for data science teams (recruiting, managing and quality-assuring) who don't come from that background themselves.

I started my company, BayesCamp, in 2017, to provide training, one-to-one coaching and consultancy. Most teaching statistics and machine learning is not as effective as it could be. I believe that we could do a lot better by helping learners to look at the concepts of models, philosophy of science, inter-personal and communication skills. I've drawn inspiration from the reform movement in stats education that Joan Garfield and George Cobb started, Mike Monteiro's code of ethics, Jenny Odell's doing nothing, and a collection of science thinkers like Charles Manski and Peter Lipton. That probably makes me an outsider teacher (analogous to an outsider artist), in that I do what I think is right, not what's popular.

I'm currently working on a few related projects in Bayesian meta-analysis: combining evidence on exercise for osteoarthritis in collaboration with St George's Medical School, building a network of people on the back of a course I gave at the Royal Statistical Society, and writing a book on the subject.

I am currently consulting or running training with: Médecins sans Frontières, the Royal College of Paediatrics and Child Health, Timberlake Consultants, Understanding ModernGov, Exploristics and the Royal Statistical Society.

I will be talking with Rick Hood at the International Web Workshop on Computational Economics and Econometrics on 2 July on the topic of Evidence and public health policy in complex systems: a view from the UK post-Covid-19.

I do a little methodological work now and then, which involves non-parametric Bayesian updating (paper under review), continuous-time Markov process samplers, ensembles, and high-dimensional rotations. I am a Stan developer, maintaining the StataStan interface and working on interactive browser outputs.

I have often worked on clinical audit and performance indicators, evidence synthesis, health services research, project management and communication of complex findings.

You can get in touch through the links on the bar above, although I especially like real-life letters and postcards: BayesCamp, 80 High Street (Unit 83), Winchester, SO23 9AT, UK.

Photo of Robert wearing his old school Stan software t-shirt Cover of data visualisation book

Cover art by Jill Pelto. jillpelto.com

Here are a few upcoming courses that I am teaching that might interest you.

An introduction to model building techniques in Stata, online, 23 June 2021
timberlake.co.uk/courses/introduction-model-building-techniques-stata.html
Ideal for anyone who uses Stata and wants to advance their skills in modelling data (supervised learning).

Advanced Statistical Analysis, online, 23 September 2021
moderngov.com
If you work in analytics in the public or charitable sector, and want to explore the more advanced techniques from statistics and machine learning, this one day introduction and overview will help. We will talk about the theory that underpins building a good analysis and critiquing it. We will also put this into practice in simple workshops using the open-source langauge R as an example, and talk about the managerial and organisational considerations.

Introduction to Bayesian Analysis using Stan, online, 6-7 July 2021
rss.org.uk
Aimed at beginners in Bayes, this course will get you up and running in using Stan for flexible, fast modelling.

Bayesian Meta-analysis, online, 14-15 October 2021
rss.org.uk
This course is for everyone who has grown tired of shoehorning their complex evidence base into a one-size-fits-all meta-analysis. Instead, you can construct a bespoke model that captures non-linearity, non-normality, biases, missing statistics and more. We assume you have encountered the basics of Bayes, perhaps having previously tried out WinBUGS / JAGS / brms. We will introduce and use Stan, but the ideas you acquire can be applied to any Bayesian software.

If there are statistical topics that would interest you and your teams, send me a message at robert@bayescamp.com and let's see if we can organise a training course that would cover it.

Uno meme joke. Check your model assumptions or draw 25.