Numpy ( 3 )

NUMPY ARRAYS: CONCATENATING, FLATTENING AND ADDING DIMENSIONS

 

Flatten VS Ravel

FLATTEN

import numpy as np
A = np.array([[[ 0, 1],
                [ 2, 3],
                [ 4, 5],
                [ 6, 7]],
                [[ 8, 9],
                [10, 11],
                [12, 13],
                [14, 15]],
                [[16, 17],
                [18, 19],
                [20, 21],
                [22, 23]]])
A.shape

(3, 4, 2)

Flatten_X = A.flatten()
Flatten_X

array([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23])

 

RAVEL

Ravel_X = A.ravel()
Ravel_X

array([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23])

 

Flatten vs Ravel

import  numpy as np
a = np.random.randint(0,10, (2,3))
a

array([[8, 9, 0], [1, 5, 8]])

print(f'변경전 {a}')
b = a.flatten()
b[0] = 10

print(f'변경후 {a}')
print(f'b : {b}')

변경전 [[8 9 0] [1 5 8]] 변경후 [[8 9 0] [1 5 8]] b : [10 9 0 1 5 8]

print(f'변경전 {a}')
b = a.ravel()
b[0] = 10

print(f'변경후 {a}')
print(f'b : {b}')

변경전 [[8 9 0] [1 5 8]] 변경후 [[10 9 0] [ 1 5 8]] b : [10 9 0 1 5 8]

⇒ ravel 은 b를 바꿔주면 a도 같이 바뀐다. 공유하고 싶지 않을때는 flatten

 

numpy.reshape()

• numpy.reshape(a, newshape, order = 'C')

arr = np.array([[0, 1],
               [2, 3],
               [4, 5]])
np.reshape(arr, (2,3))

array([[0, 1, 2], [3, 4, 5]])

arr.reshape(2,3)

array([[0, 1, 2], [3, 4, 5]])

arr_data.reshape(-1,1)

array([[0], [1], [2], [3], [4], [5]])

arr_data.reshape(1,-1)

array([[0, 1, 2, 3, 4, 5]])

X = np.array(range(24))
X

array([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23])

X.reshape((3,1,8))

array([[[ 0, 1, 2, 3, 4, 5, 6, 7]],

   [[ 8,  9, 10, 11, 12, 13, 14, 15]],
   [[16, 17, 18, 19, 20, 21, 22, 23]]])

 

 

CONCATENATING ARRAYS - 배열 연결

• numpy.concatenate((a1, a2, ...), axis=0, out=None, dtype=None, casting="same_kind")

x = np.array([11,22])
y = np.array([18,7,6])
z = np.array([1,3,5])

np.concatenate((x,y,z))

array([11, 22, 18, 7, 6, 1, 3, 5])

x = np.array(range(24))
x = x.reshape((3,4,2))
y = np.array(range(100,124))
y = y.reshape((3,4,2))

np.concatenate((x,y), axis = 0)

array([[[ 0, 1],

   [ 2, 3],

   [ 4, 5],

   [ 6, 7]],

 

   [[  8,   9],
    [ 10,  11],
    [ 12,  13],
    [ 14,  15]],

   [[ 16,  17],
    [ 18,  19],
    [ 20,  21],
    [ 22,  23]],

   [[100, 101],
    [102, 103],
    [104, 105],
    [106, 107]],

   [[108, 109],
    [110, 111],
    [112, 113],
    [114, 115]],

   [[116, 117],
    [118, 119],
    [120, 121],
    [122, 123]]])

np.concatenate((x,y), axis = 1)

array([[[ 0, 1],

   [ 2, 3],

   [ 4, 5],

   [ 6, 7],

   [100, 101],

   [102, 103],

   [104, 105],

   [106, 107]],

 

   [[  8,   9],
    [ 10,  11],
    [ 12,  13],
    [ 14,  15],
    [108, 109],
    [110, 111],
    [112, 113],
    [114, 115]],

   [[ 16,  17],
    [ 18,  19],
    [ 20,  21],
    [ 22,  23],
    [116, 117],
    [118, 119],
    [120, 121],
    [122, 123]]])

 

 

ADDING NEW DIMENSIONS - 차원 추가

x = np.array([2,5,18,14,4])
y = x[:, np.newaxis]
print(y)

[[ 2]

 [ 5]

 [18]

 [14]

 [ 4]]

 

VECTOR STACKING - 벡터 쌓기

A = np.array([3, 4, 5])
B = np.array([1,9,0])
np.column_stack((A,B))

array([[3, 1],

   [4, 9],

   [5, 0]])

np.row_stack((A,B))

array([[3, 4, 5],

   [1, 9, 0]])

 

BOOLEAN INDEXING - 참/거짓 인덱싱

x = np.array([2,5,18,14,18,18,4])
x == 18

array([False, False, True, False, True, True, False])

 

FANCY INDEXING

C = np.array([123,188,190,99,77,88,100])
A = np.array([4,7,2,8,6,9,5])
A<=5

array([ True, False, True, False, False, False, True])

C[[True, False,  True, False, False, False,  True]]

array([123, 190, 100])

 

WEIGHTED RANDOM CHOICE WITH NUMPY - 무작위 가중치 부여

from numpy.random import choice
professions = ["scientist",
"philosopher",
"engineer",
"priest",
"programmer"]
probabilities = [0.2, 0.05, 0.3, 0.15, 0.3]
choice(professions, p=probabilities)

'programmer’

programming_languages = ["Java", "C", "Python", "C++"]
weights = np.array([16, 15.2, 10, 6.1])
# normalization
weights /= sum(weights)
print(weights)
for i in range(10):
    print(choice(programming_languages, p=weights))

[0.33826638 0.32135307 0.21141649 0.12896406]

Java

Java

C++

C

Java

Java

Python

C

Java

Python