These are some simple/stupid machine learning hacks, to analyze the kepler mission data. The kepler mission intends to identify Earth-sized planets, by observing repeated transits of a planet in front of their stars, and measuring the resulting brightness reduction. The code here has some dip detection methods, and to run regression/classification on the formatted data.

The original Kepler mission data can be found here : http://archive.stsci.edu/kepler/publiclightcurves.html or here : http://kepler.nasa.gov/Science/ForScientists/dataarchive/

There are also some modules written for PyKE (http://keplergo.arc.nasa.gov/PyKE.shtml) distribution, a framework for kepler data analysis.

dipfinder/data - The original data released is in the FITS format. The example data in this directory has been formatteed into plaintext column data.

dipfinder/dipfinder.py - A dip detection program. Has the options to choose the algorithm/k-window/threshold multiplier etc.

classification/plot_oneclass.py - A one class unsupervised classfication, uses scikits.learn

pyke_modules/kepregr.py - A PyKE module to perform a k-NN regression

pyke_modules/kepdip.py - A PyKE module to perform dip detection

sv_regression/keptest.py - SV Regression, initially written as a PyKE module. Requires PyRAF/PyKE to launch