Numpy Cheat Sheet printable pdf download

SLICING (INDEXING/SUBSETTING)

)

umPy

umerical

ython

Numpy Cheat Sheet

5. Boolean arrays methods

Setting data with assignment :

Count # of ‘Trues’

(ndarray1

> 0).sum()

< 0] = 0 *

ndarray1[ndarray1

ython

ackage

in boolean array

If at least one

ndarray1.any()

If ndarray1 is two-dimensions, ndarray1 < 0

: a

reated

rianne

olton and

ean

hen

value is ‘True’

creates a two-dimensional boolean array.

If all values are

ndarray1.all()

)

‘True’

COMMON OPERATIONS

umPy

umerical

ython

1. Transposing

Note: These methods also work with non-boolean

Default data type is ‘np.float64’. This is

• A special form of reshaping which returns a ‘view’

arrays, where non-zero elements evaluate to True.

What is NumPy?

equivalent to Python’s float type which is 8

on the underlying data without copying anything.

Foundation package for scientific computing in Python

bytes (64 bits); thus the name ‘float64’.

6. Sorting

ndarray1.transpose()

Why NumPy?

If casting were to fail for some reason,

***

‘TypeError’ will be raised.

ndarray1.T

Inplace sorting

ndarray1.sort()

• Numpy ‘ndarray’ is a much more efficient way

of storing and manipulating “numerical data”

ndarray1.swapaxes(0, 1)

than the built-in Python data structures.

Return a sorted

sorted1 =

SLICING (INDEXING/SUBSETTING)

np.sort(ndarray1)

2. Vectorized wrappers (for functions that

copy instead of

• Libraries written in lower-level languages, such

take scalar values)

• Slicing (i.e.

) is a ‘view’ on

as C, can operate on data stored in Numpy

ndarray1[2:6]

inplace

the original array. Data is NOT copied. Any

‘ndarray’ without copying any data.

•

works on only a

math.sqrt()

scalar

modifications (i.e.

) to the

ndarray1[2:6] = 8

7. Set methods

N-DIMENSIONAL ARRAY (NDARRAY)

# any sequence (list,

np.sqrt(seq1)

‘view’ will be reflected in the original array.

ndarray, etc) to return a ndarray

What is NdArray?

Return sorted

np.unique(ndarray1)

• Instead of a ‘view’, explicit copy of slicing via :

unique values

3. Vectorized expressions

Fast and space-efficient multidimensional array

ndarray1[2:6].copy()

(container for homogeneous data) providing vectorized

Test membership

resultBooleanArray =

•

is a vectorized version

np.where(cond, x, y)

np.in1d(ndarray1, [2,

arithmetic operations

of ndarray1 values

of the expression ‘x if condition else y’

• Multidimensional array indexing notation :

3, 6])

in [2, 3, 6]

np.array(seq1)

Create NdArray

ndarray1[0][2]

ndarray1[0, 2]

np.where([True, False], [1, 2],

[2, 3]) => ndarray (1, 3)

# seq1 - is any sequence like object,

• Other set methods :

intersect1d()

union1d()

i.e. [1, 2, 3]

setdiff1d()

setxor1d()

* Boolean indexing :

1, np.zeros(10)

Create Special

• Common Usages :

NdArray

# one dimensional ndarray with 10

8. Random number generation (np.random)

ndarray1[(names == ‘Bob’) | (names ==

elements of value 0

np.where(matrixArray

> 0, 1, -1)

‘Will’), 2:]

• Supplements the built-in Python random * with

np.ones(2,

=> a new array (same shape) of 1 or -1 values

# ‘2:’ means select from 3rd column on

functions for efficiently generating whole arrays

# two dimensional ndarray with 6

of sample values from many kinds of probability

np.where(cond, 1, 0).argmax() *

elements of value 1

=> Find the first True element

distributions.

Selecting data by boolean indexing

np.empty(3, 4,

5) *

ALWAYS creates a copy of the data.

# three dimensional ndarray of

samples = np.random.normal(size =(3, 3))

uninitialized values

can be used to find the

argmax()

The ‘and’ and ‘or’ keywords do NOT work

index of the maximum element.

4, np.eye(N) or

Example usage is find the first

with boolean arrays. Use & and |.

np.identity(N)

Python built-in random ONLY samples

element that has a “price > number”

# creates N by N identity matrix

one value at a time.

in an array of price data.

np.arange(1, 10)

NdArray version of

* Fancy indexing (aka ‘indexing using integer arrays’)

Python’s

range

4. Aggregations/Reductions Methods

Select a subset of rows in a particular order :

Get # of Dimension

ndarray1.ndim

(i.e. mean, sum, std)

Get Dimension Size

dim1size, dim2size, ..

ndarray1[ [3, 8, 4] ]

ndarray1.shape

Compute mean

ndarray1.mean()

ndarray1[ [-1, 6] ]

Get Data Type **

ndarray1.dtype

np.mean(ndarray1)

Created by Arianne Colton and Sean Chen

# negative indices select rows from the end

Compute statistics

ndarray1.mean(axis = 1)

Explicit Casting

ndarray2

= ndarray1.

astype(np.int32) ***

over axis *

Fancy indexing ALWAYS creates a

ndarray1.sum(axis = 0)

Based on content from

copy of the data.

Cannot assume empty() will return all zeros.

‘Python for Data Analysis’ by Wes McKinney

axis = 0 means column axis, 1 is row axis.

It could be garbage values.

Updated: August 18, 2016

Numpy Cheat Sheet

Related Articles

Related forms

Related Categories