├── README.md
└── xcorr.py


/README.md:
--------------------------------------------------------------------------------
 1 | # xcorr_python
 2 | Cross-correlation coefficients in Python
 3 | 
 4 | Returns coefficients (or inner product) and lags
 5 | 
 6 | This might save someone a bit of time, I could not find a standard xcorr function (like MATLAB's) in Python, which returns the *coefficients* of a cross correlation of two signals (instead of the inner product). 
 7 | 
 8 | This code is adapted from [matplotlib's xcorr function](https://matplotlib.org/examples/pylab_examples/xcorr_demo.html "matplotlib xcorr()"), I just separated the normalization from the plotting behavior.
 9 | 
10 | Calls [numpy.correlate(x,y, mode='full')](https://docs.scipy.org/doc/numpy-1.13.0/reference/generated/numpy.correlate.html "numpy.correlate()")
11 | 
12 | #### This is the necessary transformation function (inner product -> coeffs.)
13 | n = np.sqrt(np.dot(x, x) * np.dot(y, y)) 
14 | 
15 | coeffs = np.true_divide(innerproduct,n)
16 | 


--------------------------------------------------------------------------------
/xcorr.py:
--------------------------------------------------------------------------------
 1 | def xcorr(x, y, normed=True, detrend=False, maxlags=10):
 2 |     # Cross correlation of two signals of equal length
 3 |     # Returns the coefficients when normed=True
 4 |     # Returns inner products when normed=False
 5 |     # Usage: lags, c = xcorr(x,y,maxlags=len(x)-1)
 6 |     # Optional detrending e.g. mlab.detrend_mean
 7 | 
 8 |     Nx = len(x)
 9 |     if Nx != len(y):
10 |         raise ValueError('x and y must be equal length')
11 |     
12 |     if detrend:
13 |         import matplotlib.mlab as mlab
14 |         x = mlab.detrend_mean(np.asarray(x)) # can set your preferences here
15 |         y = mlab.detrend_mean(np.asarray(y))
16 |     
17 |     c = np.correlate(x, y, mode='full')
18 | 
19 |     if normed:
20 |         n = np.sqrt(np.dot(x, x) * np.dot(y, y)) # this is the transformation function
21 |         c = np.true_divide(c,n)
22 | 
23 |     if maxlags is None:
24 |         maxlags = Nx - 1
25 | 
26 |     if maxlags >= Nx or maxlags < 1:
27 |         raise ValueError('maglags must be None or strictly '
28 |                          'positive < %d' % Nx)
29 | 
30 |     lags = np.arange(-maxlags, maxlags + 1)
31 |     c = c[Nx - 1 - maxlags:Nx + maxlags]
32 |     return lags, c
33 | 


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