Note: this package is largely obsolete, because most of its content has been integrated into Statsmodels. Please see my working paper Estimating time series models by state space methods in Python: Statsmodels for more information on using Statsmodels to estimate state space models.
A collection of resources for quantitative economics in Python. Includes:
- a Python wrapper for state space models along with a fast (compiled) Kalman filter, Kalman smoother, and simulation smoother.
- integration with the Statsmodels module to allow maximum likelihood estimation of parameters in state space models, summary tables, diagnostic tests and plots, and post-estimation results.
- parameter estimation using Bayesian posterior simulation methods (either Metropolis-Hastings or Gibbs sampling - see the user guide for details, below), including integration with the PyMC module.
- built-in time series models: SARIMAX, unobserved components, VARMAX, and dynamic factor models.
- high test coverage
See http://dismalpy.github.io/installation.html for details on installation.
- The up-to-date source code is available on GitHub: http://github.com/dismalpy/dismalpy
- Source distributions and some wheels are available on PyPi: https://pypi.python.org/pypi/dismalpy/
This package has the following dependencies:
- NumPy
- SciPy >= 0.14.0
- Pandas >= 0.16.0
- Cython >= 0.20.0
- Statsmodels >= 0.8.0; note that this has not yet been released, so for the time being the development version must be installed prior to installing DismalPy.
Simplified-BSD License
Please submit bug reports to http://github.com/dismalpy/dismalpy/issues