vinci1it2000 / syncing

Synchronise and re-sample time series

syncing: Time series synchronisation library.

release:	1.0.9
date:	2023-11-15 12:10:00
repository:	https://github.com/vinci1it2000/syncing
pypi-repo:	https://pypi.org/project/syncing/
docs:	http://syncing.readthedocs.io/
wiki:	https://github.com/vinci1it2000/syncing/wiki/
download:	http://github.com/vinci1it2000/syncing/releases/
donate:	https://donorbox.org/syncing
keywords:	data, synchronisation, re-sampling
developers:
license:	EUPL 1.1+

What is syncing?

syncing is an useful library to synchronise and re-sample time series.

synchronisation is based on the fourier transform and the re-sampling is performed with a specific interpolation method.

Installation

To install it use (with root privileges):

$ pip install syncing

Or download the last git version and use (with root privileges):

$ python setup.py install

Install extras

Some additional functionality is enabled installing the following extras:

• cli: enables the command line interface.
• plot: enables to plot the model process and its workflow.
• dev: installs all libraries plus the development libraries.

To install syncing and all extras (except development libraries), do:

$ pip install syncing[all]

Synchronising Laboratory Data

This example shows how to synchronise two data-sets obd and dyno (respectively they are the On-Board Diagnostics of a vehicle and Chassis dynamometer) with a reference signal ref. To achieve this we use the model syncing model to visualize the model:

.. dispatcher:: model
    :opt: graph_attr={'ratio': '1'}
    :code:

    >>> from syncing.model import dsp
    >>> model = dsp.register()
    >>> model.plot(view=False)
    SiteMap(...)

Tip

You can explore the diagram by clicking on it.

First of all, we generate synthetically the data-sets to feed the model:

.. plot::
    :include-source:

    >>> import numpy as np
    >>> data_sets = {}
    >>> time = np.arange(0, 150, .1)
    >>> velocity = (1 + np.sin(time / 10)) * 60
    >>> data_sets['ref'] = dict(
    ...     time=time,                                               # [10 Hz]
    ...     velocity=velocity / 3.6                                  # [m/s]
    ... )
    >>> data_sets['obd'] = dict(
    ...     time=time[::10] + 12,                                    # 1 Hz
    ...     velocity=velocity[::10] + np.random.normal(0, 5, 150),   # [km/h]
    ...     engine_rpm=np.maximum(
    ...         np.random.normal(velocity[::10] * 3 + 600, 5), 800
    ...     )                                                        # [RPM]
    ... )
    >>> data_sets['dyno'] = dict(
    ...     time=time + 6.66,                                        # 10 Hz
    ...     velocity=velocity + np.random.normal(0, 1, 1500)         # [km/h]
    ... )

    To synchronise the data-sets and plot the workflow:

    .. dispatcher:: sol
        :opt: workflow=True, graph_attr={'ratio': '1'}
        :code:

        >>> from syncing.model import dsp
        >>> sol = dsp(dict(
        ...     data=data_sets, x_label='time', y_label='velocity',
        ...     reference_name='ref', interpolation_method='cubic'
        ... ))
        >>> sol.plot(view=False)
        SiteMap(...)

    Finally, we can analyze the time shifts and the synchronised and re-sampled
    data-sets:

    >>> import pandas as pd
    >>> import schedula as sh
    >>> pd.DataFrame(sol['shifts'], index=[0])  # doctest: +SKIP
         obd  dyno
    ...
    >>> df = pd.DataFrame(dict(sh.stack_nested_keys(sol['resampled'])))
    >>> df.columns = df.columns.map('/'.join)
    >>> df['ref/velocity'] *= 3.6
    >>> ax = df.set_index('ref/time').plot(secondary_y='obd/engine_rpm')
    >>> ax.set_ylabel('[km/h]'); ax.right_ax.set_ylabel('[RPM]')
    Text(...)

syncing's Issues

Correct `bs4` dependency in `setup.py` is a dummy package linking to `beautifulsoup4`

Update to the canonical pypi name of beautifulsoup4.

Description in setup.py pasted from formulas

ENH: build as "universal" wheel

Currently packaged 'wheels' are for the specific platform and python-version,
but actually this project contains pure-python code so it can "universal".

Please the following section in setup.py:

[bdist_wheel]
universal = 1

get_long_description() in setup.py fetchs too-many dependencies on build

setup.py scripts are supposed to import as little as possible because they python-environment might not have all those libraries installed.
And by invoking sphinx-builder, it loads in build-time all libraries used by docs (matplotlib & graphviz).

• Practically it is not easy to test the project because of all those dependencies
• Conda build needs special handlng.

Alternative

use the pure-python code from pandalone to clean README.rst for PyPi.