Dear folks, here I am again after moving to a new place, going on vacation with my fellow mate (and some math books too - I really can’t stand engineers longer than three days) and even to a summer school in advanced statistics in the very north of Europe (oo) Have you watched the Olympic Games in London? No?! Well, amazing as usual. Lots of records and medals, numbers and measures. Plus I ended up on reading a paper about the predictability of gold-medal performances at the Olympics via the analysis of past results. To be honest the title sounded pretentious since the beginning. But I had to give it a chance and add it to my collection of read papers that deserve a note.
The paper I am talking about is Universality, limits and predictability of gold- medal performances at the Olympic Games** **written by Filippo Radicchi. Here the author draws a picture of future improvements through an analysis of historical results achieved by athletes who won a gold medal in the past in different disciplines of athletics. The assumption is that those results are destined to improve up to a structural/physical limit. Which is what can be observed from historical records, indeed. Unfortunately, by ignoring the improvements headed by, for instance, gear, equipment, nutrition, quality of life, achievements of the science of training etc. the model stays incomplete and the prediction might result poor. I consider poor predictions the ones that lay in a consistently large numerical range. The poorness of the prediction does not appear from the paper just for one specific reason: it goes accordingly with the assumption. I am quite skeptic about this. Moreover, as Jacques-Louis Lions said once, ”do you really want to mathematize everything?” (oo)
1. Universality, limits and predictability of gold-medal performances at the Olympic Games by Filippo Radicchi. 2. 60 sec science - New materials (podcast)
Before you go
If you enjoyed this post, you will love the newsletter of Data Science at Home. It’s my FREE digest of the best content in Artificial Intelligence, data science, predictive analytics and computer science. Subscribe!