Metadata-Version: 2.1
Name: microfilter
Version: 0.0.5
Summary: Filtering
Home-page: https://github.com/microprediction/microfilter
Author: microprediction
Author-email: pcotton@intechinvestments.com
License: MIT
Platform: UNKNOWN
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.7
Description-Content-Type: text/markdown
Requires-Dist: pandas
Requires-Dist: numpy
Requires-Dist: pytest
Requires-Dist: statsmodels
Requires-Dist: hyperopt
Requires-Dist: scipy
Requires-Dist: matplotlib
Requires-Dist: microconventions (<0.4)

# microfilter

Some ad-hoc approaches to filtering noisy data that don't appear in textbooks

![](https://i.imgur.com/b5fAtxr.png)

## Usage example

Train filter on simulated noisy data

    from microfilter.univariate.expnormdist import ExpNormDist
    from microfilter.univariate.noisysim import sim_lagged_values_and_times

    lagged_values, lagged_times = sim_lagged_values_and_times
    dist = ExpNormDist()
    dist.hyper_params['max_evals']=500
    dist.fit(lagged_values=lagged_values, lagged_times=lagged_times)
    pprint(dist.params) 
    new_value = 17.0
    dist.update(value=new_value, dt=1.0)
    pprint(dist.state) 

See https://github.com/microprediction/microfilter/blob/master/examples/plot_expnorm.py 




