Show Menu
Cheatography
  1. Home >
  2. Programming >

Numpy Python3 Cheat Sheet (DRAFT) by

Numpy cheatsheet

This is a draft cheat sheet. It is a work in progress and is not finished yet.

Operataing element by element

Operate n to all
components
 
vector + n

matrix  n

vector * n
Operation using another vector 
vector + vector 

matrix * matrix
Matrix : Transpose 
matrix.T
Matrix multip­lic­ation 
np.dot­(ma­tri­x,m­atr­ix.T)

Vectors & Matrix

Create vector from list
 
x = np.arr­ay(x)
Create vector of n zeros
 
zeros = np.zer­os(n)
Create vector of n ones
 
ones = np.ones(n)
Create vector of type float32
 
np.arr­ay(­[1,­-0.3­,2], dtype=­np.f­lo­at32)
Change type of a vector
 
vector = vector.as­typ­e(n­p.u­int32)
Create matrix of ones
 
m1 = np.one­s((­5,4))
Create matrix from other matrix shape
 
m2 = np.zer­os_­lik­e(m1)
Reshape a matrix (constant size)
 
mat=ma­t.r­esh­ape­((6,3))
 

Matrix & Vectors Operations

Sum of all components
 
np.sum­(ve­ctor)

vector.sum()
Maximum
 
np.max­(ve­ctor)

vector.max()
Minimum
 
np.min­(ve­ctor)

vector.min()
Product
 
np.pro­d(v­ector)

vector.prod()
Operation over column
 
matriu.su­m(a­xis=0)
Operation over row
 
matriu.su­m(a­xis=1)
Absolute
 
np.sum­(ma­triu)

Indexing

Especific element
 
matrix[row,column ]
Range of row / columns
 
matrix[ rI : rE, cI :cE]
Submatrix (same matrix)
 
submatrix = matrix[1:3, 1:3]
Submatrix
(Different matrix)
 
submatrix = submatrix.copy()
Boolean Indexing
Finding negative values
negatius = example < 0
Selecting negative values
exampl­e[n­ega­tius]
Changing values directly
exampl­e[e­xam­ple==0] = 0
 

Broadc­asting

Normalize a matrix by columns
 
np.random.uniform(size=(3,4))

sumes = matrix.sum(axis=0)

norm = matrix/sumes
Normalize a matrix by rows**
 
norm=(matrix.T/sumes).T

norm=matrix/sumes.reshape((3,1))
Not-a-­Number ( Nan)
Initialize as NaN
`a = np.nan
Test if is NaN 
np.isn­an(a)
Test also with vectors 
np.isn­an(­vector)

Extra

Random uniform matrix 
np.random.uniform(size=(3,4))
Another type of
random matrix
 
np.random.uniform(-1,1,size=(3,4))
Copying a matrix
 
matrix = matrix.copy()
Shape of a matrix 
matrix.shape
Data types 
dtype = np.float32

dtype = np.float64

dtype = np.int32

dtype = uint32

dtype = np.int8

dtype = np.int8

dtype = np.uint8
Importing 
import numpy as np