Introduction It’s been three years since the Casualty Actuarial Society published our research paper on Hierarchical Compartmental Reserving Models (Gesmann and Morris (2020)). Time to revisit it, as developments of the Stan language, and its interfaces such as cmdstanr and brms have progressed and simplified the treatment of differential equations. We have updated the bookdown version version of the paper to take advantage of these newer versions. This post will give another example of how to use hierarchical compartmental reserving models, but rather than working with historical claims data ..read more

This post is about the Black-Litterman (BL) model for asset allocation and the basis of my talk at the Dublin Data Science Meet-up. The original BL paper (Black and Litterman (1991)) is over 30 years old and builds on the ideas of modern portfolio theory by Harry Markowitz (Markowitz (1952)). A good introduction to the BL model is (Idzorek (2005)) or (Maggiar (2009)). I am not sure how much the model is used by investment professionals, as many of the underlying assumptions may not hold true in the real world. However, the model uses some great ideas, including Bayes’ rule, to combine differen ..read more

Version 0.7.0 of the googleVis R package has been released, adding a new function for Gantt charts. Gantt charts are helpful to illustrates a project schedule and its dependencies. Following the Google documentation the project has to be broken down into task IDs, task names, resources, start date, end dates, task duration (in milliseconds), how far the task has been completed (in percent), and finally any dependencies to other tasks IDs. Here is an example for a project to write a research paper: # Helper function daysToMilliseconds <- function(days){ days * 24 * 60 * 60 * 1000 } # Proje ..read more

It is exactly one year ago that the Casualty Actuarial Society published our research paper on Hierarchical Compartmental Reserving Models (Gesmann and Morris (2020)). One aspect we looked into was the question if the choice of modelling cumulative or incremental payment data over time matters. Many traditional reserving methods (including the chain-ladder technique) take cumulative claims triangles as an input. Plotting cumulative claims data allows us to quickly understand key data features by eye. The same observation applies to other growth data, e.g. the growth development of childre ..read more

On Sunday the Tokyo Olympics men sprint 100m final will take place. Francesc Montané reminded me in his analysis that 9 years ago I used a simple regression model to predict the winning time for the 100m men sprint final of the 2012 Olympics in London. My model predicted a winning time of 9.68s, yet Usain Bolt finished in 9.63s. For this Sunday my prediction is 9.72s, with a 50% credible interval of [9.61s, 9.85s]. Since the 2012 Olympics things changed, Usain Bolt has retired and new shoes with advance spike technology have created a little bit of a controversy. I don’t think a log-linear reg ..read more

Finally, the Insurance Data Science conference was back last week. After last year’s cancellation due to Covid-19 over 250, delegates from around the world came together on-line for the third instalment of the conference. The event kicked-off, or should we say lifted off, with a keynote by Thomas Wiecki, CEO of PyMC Labs, on Wednesday. Thomas explained how probabilistic programming can be used to assess risk and make decision in the context of insuring rocket launches. Although all talks took place on Zoom, the interaction of the audience with the speakers was live and active, particularly so ..read more

The programme and abstract booklet for the 2021 Insurance Data Science (16 - 18 June) conference (fka R in Insurance) is now out! We are very excited for a great range of speakers and topics. Previous Next     Page: / Download the Programme and Abstract Booklet Book your ticket for this three half day event by 9 June on Eventbrite. The conference fee for the Insurance Data Science conference is: Practitioners: £150 Academics and Postgraduate Students: £50 Thanks to our sponsors Swiss Re Institute and Mirai Solutions GmbH! For more information visit: https://insurancedatascience.or ..read more

This article illustrates how ordinary differential equations and multivariate observations can be modelled and fitted with the brms package (Bürkner (2017)) in R1. As an example I will use the well known Lotka-Volterra model (Lotka (1925), Volterra (1926)) that describes the predator-prey behaviour of lynxes and hares. Bob Carpenter published a detailed tutorial to implement and analyse this model in Stan and so did Richard McElreath in Statistical Rethinking 2nd Edition (McElreath (2020)). Here I will use brms as an interface to Stan. With brms I can write the model using formulas similar to ..read more

Over the last year I worked with Jake Morris on a research paper for the Casualty Actuarial Society. We are delighted to see it published: Gesmann, M., and Morris, J. “Hierarchical Compartmental Reserving Models.” Casualty Actuarial Society, CAS Research Papers, 19 Aug. 2020, https://www.casact.org/research/research-papers/Compartmental-Reserving-Models-GesmannMorris0820.pdf The paper demonstrates how one can describe the dynamics of claims processes with differential equations and probability distributions. All of this is set into a Bayesian framework that allows us to combine judgeme ..read more

..read more