PyDP is library for implementing Dirichlet Process mixture models (DPMM). The goal of PyDP is to provide a pure Python implementation of various algorithms for working DPMMs. As a design choice PyDP should have no dependencies on any libraries which are not supported by the PyPy Python interpreter.
PyDP is licensed under the GPL v3, see the LICENSE.txt file for details.
- Fixed bug in mpear
-
Added code for vector distributions
-
Added code for clustering using MPEAR
-
Fixed a bug in concentration sampler
-
Fixed log_beta function to check if parameters are <= 0 and return -inf if so
-
Changed the interface for AtomSampler to take cells instead of partitions.
-
Added global parameter updating.
-
Updated density interface to use caching.
-
Added some new proposal functions.
- Fixed error in concentration sampler due to using the wrong parameterisation of the gamma prior.
- Fixed underflow issue in precision update for Gaussian model.
-
Added code for Gaussian models.
-
Added wrapper class for DP sampler.
- Added GPL license informtation.
Installation is the standard python setup.py install
.
- None
- SymPy >= 0.7.1 - Used for some of the diagnostic tools to compute the chi-square distribution.