Bugfix release and performance improvements

This is a bugfix release which fixes a bug in the Hamliton detector and also speeds up its operation dramatically.

bugfix release

This fixes a bug were a missing peak in the Pan Tomkins detector could crash the detector. Thanks for @augelloantonio for pointing it out.
The usage example now also calculates the heartrate and shows how to convert from the sample numbers to time.

WQRS performance improvements

The WQRS detector is using now numpy array commands which speeds up its operation substantially.

Bugfix release (WQRS)

• The wqrs detector had a bug in the calculation of the length transform which resulted in poor performance. This has now been fixed.
• The template of the matched filter was converted to a python module containing two arrays. Makes it much more reliable cross platforms to load the template.

Wavelet detector fix release and detector list

The implementation of the wavelet detector was not 100% following the methods in the paper. This has been fixed. See: 94df918

The detector class has now an array with descriptions of the detectors and pointers to the methods. The usage example script show how the array can be used to offer choice of detectors.

wqrs_detector

The wqrs-detector (http://cinc.mit.edu/archives/2003/pdf/737.pdf) has been added -- thanks Yoav Nir for sending the pull request.

Matched filter update

The matched filter so far only used the templates we have provided at two sampling rates but it should really take a template as an argument so that the user can supply the template. If the user omits the template argument it still tries to load the pre-defined templates from the templates folder. However, that's more for backwards compatibility.

r_peaks = detectors.matched_filter_detector(unfiltered_ecg,"templates/template_250hz.csv")

Minor bugfix release

This release removes some unnecessary packages references such as biosppy and pathlib. Otherwise it's identical to the previous release. After half a year no bugs have been reported I bump up the version number to 1.0.0.

Stats update

Added a threshold that we need at least 20 sensitivity pairs to calc any p-value.
Made the output exactly as it's in our corresponding paper.

package bugfix release

There hasn't been any changes to the code itself but the actual package management with pypi has been tested extensively where dependencies have been fixed and other minor issues sorted.