Numpy extra

Last modified on 01 Oct 2021.

Import library
Checking
Creating
Deleting

Import library

import numpy as np

Checking

Isinstance

Check if an object is an instance numpy array?

type(MyObect).__module__ == np.__name__

Comparing

Compare 2 dicts of multiple objects:

# 2 dicts of multiple objects
def _compare_two_dict(dct_1, dct_2):
    return np.array([(lambda key: (dct_1[key] == dct_2[key]).all()
       if type(dct_1[key]).__module__ == np.__name__
       else dct_1[key] == dct2[key])(key)
       for key in dct_1]).all()

# 2 numpy arrays containing `np.nan`
def nan_equal(a,b):
    try:
        np.testing.assert_equal(a,b)
    except AssertionError:
        return False
    return True

# if using in a pytest
# instead of `assert np.testing`,
# just use
np.testing.assert_equal(a,b)

Count 0, NaNs

# count nans
arr = [1,2,3,4, np.nan,np.nan, 0,0,0]
np.isnan(arr).sum() # 2

# count non-nans
arr = [1,2,3,4, np.nan,np.nan, 0,0,0]
np.count_nonzero(~np.isnan(arr)) # 7

# count non-zeros
arr = [1,2,3,4, np.nan,np.nan, 0,0,0]
np.count_nonzero(arr) # 6

Creating

Random numbers

# random int between 0, 5
np.random.randint(5)

# random int between [1, 100]
np.random.randint(1, 100 + 1)

# random array of int between 1, 5
np.random.randint(1,5,size=(2,3))

# random array of int between 0, 3
np.random.randint(3,size=(2,3))

# random float number between [0, 1)
np.random.random()

# random float number between [a, b)
(b - a)*np.random.random() + a

# array of random between [0, 1)
np.random.random_sample((5,)) # size: 5x1

# array of random between (a, b)
(b - a)*np.random.random_sample((5,1)) + a

Equal size

Create evenly spaced numbers over a specified interval^[ref]

x = np.linspace(0, 3.5, num=20) # default num = 50

Range of int numbers

np.arange(0, 5)
np.arange(0, 31, 5)

array([0, 1, 2, 3, 4])
array([ 0,  5, 10, 15, 20, 25, 30])

Indexes and values in other arrays

Create an array from nested arrays. Values in array_2 are indexes in array_1 and we create a new array take values in array_1 which is corresponding to its indexes showed in array_2.

array_1 = np.array([ [0,0,0], [1,1,1], [2,2,2] ])
array_2 = np.array([1,0,2,1,0,2,1]) # indexes in array_1
array_3 = array_1[array_2]

print(array_1)
print(array_2)
print(array_3)

[[0 0 0]
 [1 1 1]
 [2 2 2]]

[1 0 2 1 0]

[[1 1 1]
 [0 0 0]
 [2 2 2]
 [1 1 1]
 [0 0 0]]

Array of NaN values

# single array
np.repeat(np.nan, 5)

# multi dimensional arrays
a = np.empty((2,3))
a[:] = np.nan
# other way
np.repeat([np.repeat(np.nan, 3)], 2, axis=0)

array([nan, nan, nan, nan, nan])

array([[nan, nan, nan],
       [nan, nan, nan]])

Repeated values

np.full((5,), 50) # 5x1 of 5

Deleting

# DELETE BY POSITIONS
arr = np.arange(6)
np.delete(arr, [3,4])

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

# DELETE BY VALUES
a = np.array([1, 2, 3, 4, 5, 6])
np.delete(a, np.where(a == 3))

b = np.array([1, 2, 3, 'a'])
np.delete(a, np.where(b == 'a'))

# OR
a[a!=3]
b[b!='a']

array([1, 2, 4, 5, 6])

array(['1', '2', '3'], dtype='<U21')

# Remove 'NaT' from an array
Y = np.array([600000, 300000, 'NaT'],
              dtype='timedelta64[ns]')
Y[~np.isnat(Y)]

array([600000, 300000],
       dtype='timedelta64[ns]')