The latest prediction competition at Kaggle is literally "out of this world": the goal is to quantify the shape of 2-D images of galaxies from a simulated telescope, to test models for how invisible dark matter in the Universe distorts the images through gravitational lensing (as shown in the image below; see the FAQ for more details). If you're thinking about tackling this in R, the pixmap package will be handy for importing the images, and wavelet analysis might be a good way of dealing with the noise issues. The winner will receive an trip to JPL in Pasadena, Calfornia to attend the GREAT10 challenge workshop "Image Analysis for Cosmology".
Kaggle: Mapping Dark Matter
Comments