Hint is a hyperinterval finder ============================== Copyright (c) 2011-2012 Jouke Witteveen <jwitteve@liacs.nl>. All rights reserved. About ----- Hint takes a numerical database, samples it and using compression techniques determines an interesting hyperinterval (multi-dimensional interval). The data is fitted to the following interval form: [-A-)[-B-](-A-] where a codelength based on a uniform density on A and B is calculated. The choice of the splits as model parameters is taken with a uniform prior, which means its cost can be discarded without loss of generality. This is untrue for comparison to the single uniform density coding [-A-]. Think of the middle interval [-B-] as a luckiness region. Dependencies ------------ - Python 3 (as 'python') Usage ----- Usage information of the bare program is available through ./source/hint.py --help The wrappers need access to the main program. One way to achieve that is by calling them as PYTHONPATH=./source/ ./test/source/<wrapper>.py <arguments>