Bootstrap particle filter for epidemic forecasting¶
Welcome to the pypfilt
documentation.
This package implements a bootstrap particle filter that can be used for
recursive Bayesian estimation and forecasting.
Quick-start¶
Here’s a brief example of how to use pypfilt
directly:
import datetime
import pypfilt
params = ... # Simulation parameters, model parameters, etc.
obs = ... # A list of observation streams.
summary = ... # An object that calculates summary statistics.
start = datetime.datetime(2015, 1, 1)
until = datetime.datetime(2015, 12, 31)
outfile = 'forecasts.hdf5'
# Generate forecasts and save them to disk (in HDF5 format).
forecasts = pypfilt.forecast(params, start, until, obs, summary, outfile)
See Getting Started for a more detailed example.
Installation¶
Clone the pypfilt
repository and run:
python setup.py install
If you don’t have admin rights, install the package locally:
python setup.py install --user