Metadata-Version: 2.1
Name: trouve
Version: 0.6.0
Summary: Event detection and transformation for time-series data
Home-page: https://github.com/rwhitt2049/trouve
Author: Ry Whittington
Author-email: rwhitt2049@gmail.com
License: MIT
Keywords: time_series,timeseries,iot,sensor
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: Microsoft :: Windows
Classifier: Operating System :: POSIX
Classifier: Programming Language :: Python :: 3.3
Classifier: Programming Language :: Python :: 3.4
Classifier: Programming Language :: Python :: 3.5
Classifier: Programming Language :: Python :: 3.6
Classifier: Topic :: Scientific/Engineering
Provides-Extra: develop
License-File: LICENSE.txt

Trouve

======



A package to search for events in time-series data that match a boolean condition. Various

transformation functions are built in to filter and alter events.



See Trouve's documentation at https://trouve.readthedocs.io



Install

-------

``trouve`` is on the Python Package Index (PyPI):



::



   pip install trouve



Dependencies

------------



* ``numpy``



* ``pandas``



* ``toolz``



Example

-------



This finds events in a short sample of 1Hz, time-series data and filters out events based

on their duration



.. code-block:: python



   >>> import numpy as np

   >>> from trouve import find_events

   >>> import trouve.transformations as tt

   >>> x = np.array([1, 2, 2, 2, 0, 1, 2, 0, 2, 2])

   >>> period = 1 # period = 1 / sample_rate

   >>> duration_filter = tt.filter_durations(2, 3)

   >>> events = find_events(x == 2, duration_filter, period=1)

   >>> len(events)

   2

   >>> events.as_array()

   array([ 0.,  1.,  1.,  1.,  0.,  0.,  0.,  0.,  1.,  1.])

