miniature-adventure

A photometric supernova classifier using a data driven approach.

Trained and tested on simulated dataset http://sdssdp62.fnal.gov/sdsssn/SIMGEN_PUBLIC/SIMGEN_PUBLIC_DES.tar.gz

Developed in Python and R languages as PhD project, and inspired to paper from Richards, J.W. et. al (2012).

Thesis will be available soon after defense in fall 2015.

Processing steps:

1. Correction for astrophysical effects
2. Interpolation using Gaussian processes
3. Parameter extraction performed by diffusion maps
4. Classification model built with random forest algorithm

Important notes

• Interpolation is performed using Python package GPy, available at https://github.com/SheffieldML/GPy
• Parameter extraction is performed by R package diffusionMap
• The classification model is built using R package randomForests