https://github.com/felixpatzelt/scorr

Fast and flexible two- and three-point correlation analysis for time series using spectral methods.
https://github.com/felixpatzelt/scorr

correlation correlations python scientific spectral-methods time-series

Last synced: 23 days ago
JSON representation

Fast and flexible two- and three-point correlation analysis for time series using spectral methods.

• Host: GitHub
• URL: https://github.com/felixpatzelt/scorr
• Owner: felixpatzelt
• License: mit
• Created: 2017-09-28T23:42:53.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2019-01-07T23:30:22.000Z (almost 6 years ago)
• Last Synced: 2024-11-23T21:16:29.455Z (about 1 month ago)
• Topics: correlation, correlations, python, scientific, spectral-methods, time-series
• Language: Python
• Size: 5.78 MB
• Stars: 10
• Watchers: 2
• Forks: 4
• Open Issues: 0
• Metadata Files:
  • Readme: README.rst
  • Changelog: CHANGELOG.rst
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
scorr

=====

Fast two- and three-point correlation analysis for time series

using spectral methods.

The calculations are FFT-based for optimal performance and offer many options 

for normalisation, mean removal, averaging, and zero-padding. In particular, 

averaging over pandas groups of different sizes (e.g. different days) is 

supported.

    

======================  ======================================================

Function                Synopsis

======================  ======================================================

acorr                   Calculate autocorrelation or autocovariance

acorr_grouped_df        Calculate acorr for pandas groups and average

corr_mat                Convert correlation vector to matrix

fft2x                   Calculate cross-bispectrum

fftcrop                 Return cropped fft or correlation

get_nfft                Find a good FFT segment size for pandas groups of 

                        different sizes 

padded_x3corr_norm      Normalise and debias three-point cross-correlations

padded_xcorr_norm       Normalise and debias two-point cross-correlations

x3corr                  Calculate three-point cross-correlation matrix

x3corr_grouped_df       Calculate x3corr for pandas groups and average

xcorr                   Calculate two-point cross-correlation or covariance

xcorr_grouped_df        Calculate xcorr for pandas groups and average

xcorrshift              Convert xcorr output so lag zero is centered

======================  ======================================================

The algorithms to calculate three-point correlations and details of daily

averaging over high-frequency trading data are described in:

	

    Patzelt, F. and Bouchaud, J-P. (2017):

    Nonlinear price impact from linear models. 

    Journal of Statistical Mechanics: Theory and Experiment, 12, 123404. 

    Preprint at `arXiv:1708.02411 /arxiv.org/abs/1708.02411>`_.

More code from the same publication is released in the `priceprop 

`_ package. 

Please find further 

explanations in the docstrings and in the examples 

directory. 

Installation

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

	pip install scorr

	

Dependencies (automatically installed)

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

    - Python 2.7 or 3.6

    - NumPy

    - SciPy

    - Pandas    

   

    

Optional Dependencies required only for the examples (pip installable)

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

    - Jupyter

    - Matplotlib

    - colorednoise