My Story: Statistical Practice in Epidemiology using R

June 2017 I attended the course “Statistical practice in Epidemiology using R”. The course was held in Tartu, Estonia, which is roughly a 2.5 hour bus ride from Tallinn airport. Tartu is a beautiful and almost mediaeval town and sort of the Estonian equivalent to Oxford. Moreover, Estonia will most likely become one of the most interesting countries in the world for epidemiology as they have excellent patient registries and have genotyped a high share of the population.

My background for attending the course was that I had learned how to use R (an alternative to SPSS and STATA) basically by myself, but only for small and very specific tasks. In practice this meant that every time I wanted to do something new, it would take me two weeks to figure out how to do it or I would end up doing stuff in a very bizarre and time consuming way. In addition, it was clear to me that I wasn’t even close to utilizing the full potential of R. So, when I spotted the course at the homepages of EPINOR, I thought I could give it a go.

The course has been held almost annually since 2000 and the faculty includes people who have contributed in the design of R and some of the core packages in addition to being excellent and high-impact research epidemiologists. The first two days, was more or less dedicated to the R-program itself, and getting a feeling for basic data wrangling and statistics. As some of you might know, for example getting your data into R for the first time can feel like passing Cerebrus on your way to Hades. From then on, the course shifted gears and it became gradually more theoretically advanced, statistically speaking.

In my opinion it was a good course. I finally had a proper formal introduction to R, which was urgently needed, and I learned how to use R in a more productive way. Moreover, I was introduced to a range of methods and concepts that I had heard about, but didn’t know anything about, for example spline modelling. And as always, it was nice to discuss methodology and research with experts from other fields of research in an extremely informal setting.

If there is anything negative to say about the course, it is that some of the lectures seemed sort of outdated. For example, we used the standard plotting of R, but I think most of those who are starting with R nowadays, just use ggplot2 instead as it is both more intuitive and easier to work with.

Would I recommend the corse? Yeah, especially if you’re a novice in R and have a genuine interest in statistical methodology. The week I spent in Tartu has already paid off, as I have saved a lot of time in my use of R since I came back.

By the way: In 2018, the course is most likely going to be held in Lyon, France.

For complete program (and enlightening slides), see: http://bendixcarstensen.com/SPE/

Tor-Arne Hegvik

 



Page administrator:
Last updated: 20.10.2017 10:58