admin's picture

Background

Purpose of the GEOSTAT Spatial Prediction Competition Game (the GEOSTAT hackaton) is to provide an opportunity to students to use all obtained knowledge during the course for solving a real-life problems. This typically implies producing predictions for a given data set (not previously published). It is the ultimate test of your GEOgraphy-STATistics skills. The winners are announced on the last day of the Summer School and symbolic awards are given... and of course stay fameous forever!

Standard rules

GEOSTAT Spatial Prediction Competition Game follows some standard minimum rules:

  • The winner is decided purely by numbers. Usually the smallest RMSE at validation points i.e. who ever generates predictions closest to the actual values.
  • The Game opens on Monday and closes on Friday. The deadlines are followed strickly (in seconds).
  • Only participants of the active GEOSTAT Summer School (including the lecturers) can participate.
  • The Game participants have complete freedom to use any information and/or method they can find to generate predictions at validation points (they should, however, be able to reproduce the results in front of the class, including the steps used to prepare the data).
  • The validation and training data sets are usually about the same size.
  • Participants can compete individually, or in in group of 2 max (in which case the winning group will split the award).
  • The award includes: a bon to obtain books on Amazon + photograph on the website (i.e. the eternal fame).

Previous winners

GEOSTAT 2017 Split


Best predictions: Barry Rowlingson


Mapping / visualization: Carl Reynolds

 

GEOSTAT 2016 Albacete


Best predictions: Jose Antonio Vazquez Moris & Alzira Ramos


Mapping / visualization: Antonio Jesus Pérez-Luque

 

GEOSTAT 2015 Lancaster


Best predictions: Jean Gaudart


Mapping / visualization: Robin Lovelace

 

GEOSTAT 2014 Bergen


Best predictions: Hanna Meyer