python_cheatsheet.md

python-cheatsheet

materials collected for anaconda-python 3.5

general

• to run a script from within a prompt, use exec(open('my_script.py').read())

numpy

np.zeros((2, 3, 4)) | create a three-dimensional numpy-array, filled with zeros my_3d_array.sum(axis = 1) | sum a multi-dimensional np-array over the first axis sum(my_1d_array) | an alternative way of doing summation 'np.sum(my_matrix[:, 2]) | sum over all rows in the third column my_2d_array[0] | pick the 0-entries along the first-dimension my_2d_array[:, 0] | pick the 0-entries along the second-dimension np.outer(v1, v2) | computes the outer-product-matrix of two vectors my_arr.shape | returns the dimensions of an tensor using numpy to import data from .dta-example

    movies = np.loadtxt(
      filename,
      dtype={
        'names': ('movieid', 'moviename'),
        'formats': ('int32', 'S100')},
      delimiter='\t')

np.random.randint(1, 6 + 1, size = 10) | generate an array of 10 random integers

basic operations

[some_func(x) for x in my_arr] | apply a function to every element in my_arr

plotting

• simple scatterplot:

import matplotlib.pyplot
import pylab
matplotlib.pyplot.scatter([1, 2], [1, 4])
matplotlib.pyplot.show()

• it is not necessary to provide y-values explicitely:

plt.figure()
plt.plot(x, np.log(x))
plt.plot(x, x - 1)
plt.xlabel('x')
plt.legend(['ln(x)', 'x - 1'], loc=4)
plt.show()

sources

• computational probability and inference-course, edX
• stackoverflow, of course