Quick Reference of NumPy and SciPy

Last modified: Jun, 2014

General info

• NumPy functions on sorting, searching and counting

Linear regression

Both NumPy and SciPy have functions for simple linear regression (see post). It is more intuitive to do in SciPy as follows:

from scipy import stats

x = [1, 2, 3, 4]
y = [2, 4, 7, 9]

slope, intercept, r_value, p_value, std_err = stats.linregress(x,y)

The meaning of each output is self-explanatory in the code above.

Find indices of an array

The “find” function in Matlab is very handy, but it doesn’t have a direct clone in NumPy.

• Most commonly, function numpy.where is the closest one to the “find” in Matlab:

import numpy as np 

a = np.array([6,7,8,9])
print np.where((a > 6) & (a <9))

will return a tuple “(array([1, 2]),)” which is the array of matched indices.

• Or, a custom function mimicking the find in Matlab should great.

Related functions on sorting, searching and counting are here.

Find the size of a matrix

This is equivalent to the “size” function in Matlab:

(m, n) = mtx.shape

where “mtx” is a matrix.

However, there is a caveat: if the matrix is an 1D array, then the returned tuple may miss the corresponding “1”. For example:

import numpy as np 
a = np.linspace(1,10, 10)
print a.reshape(5,2)[:,0].shape

may just return (5,) instead of (5,1). This is the difference from the “size” function in Matlab.

NumPy for Matlab Users

http://wiki.scipy.org/NumPy_for_Matlab_Users

Read Text Tables in Python

http://penandpants.com/2012/03/09/reading-text-tables-with-python/