Metadata-Version: 2.0
Name: ledge
Version: 0.0.5
Summary: Hedging algorithms with lag
Home-page: https://github.com/reichlab/ledge
Author: Abhinav Tushar
Author-email: lepisma@fastmail.com
License: GPLv3
Platform: UNKNOWN
Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3)
Classifier: Natural Language :: English
Classifier: Programming Language :: Python
Classifier: Programming Language :: Python :: 3 :: Only
Requires-Dist: numpy
Requires-Dist: pandas
Requires-Dist: xarray

# ledge

[![Build
Status](https://img.shields.io/travis/reichlab/ledge.svg?style=flat-square)](https://travis-ci.org/reichlab/ledge) [![Pypi](https://img.shields.io/pypi/v/ledge.svg?style=flat-square)](https://pypi.python.org/pypi/ledge)

Hedging algorithms with lag.

Besides the regular setting of 'prediction with expert advice', we are
interested in working with truth values with _lags_. This results in a partially
observed truth value for the present (and the past) at each step. In a discrete
time setting with delays in accurate characterization of _final_ truth, a truth
value is specified by:

1. `timepoint`: The time itself
2. `lag`: Time the value was revealed - `timepoint`

## Usage

`ledge` is a composed from a bunch of types and functions. It works with
[DataArrays](http://xarray.pydata.org/en/stable/generated/xarray.DataArray.html#xarray.DataArray)
for model _predictions_, _losses_ and _truths_.

There is no single model and the user is supposed to _compose_ a model using the
components in here. Each module is written as a literate
[org-mode](https://orgmode.org/) file and contains usage documentation for the
containing functions. Start with the main file `./ledge/README.org`.


