Working with NDArray

Working with NDArray

Get shape of array

We can get shape of an array by accessing 'shape' property.

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4)])
    print array.shape
    print len(array.shape)
    print array.shape[0]
    print array.shape[1]

if __name__ == "__main__":
    test_run()

Here we have printed out the shape of the array.

(2, 3)
2
2
3

Get size of array

Size of array is returned as sum of all cells in the n-dimension array.

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4)])
    print array.size

if __name__ == "__main__":
    test_run()

For our example, which contains 6 numbers, it prints out 6.

6

Type of array

NumPy arrays are homogenous (all the cells must have the same type) and we can get that type for an array.

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4)])
    print array.dtype

if __name__ == "__main__":
    test_run()

Here is the type of the array.

int64

Make random numbers the same for every execution

We can set seed for random number generation. Which will make sure that the random numbers are generated always the same.

import numpy as np

def test_run():
    np.random.seed(100)
    array = np.random.randint(0, 10, size=(3,4))
    print array

if __name__ == "__main__":
    test_run()

So, everytime we run this code, we get the same result.

[[8 8 3 7]
 [7 0 4 2]
 [5 2 2 2]]

Sum all elements in array

import numpy as np

def test_run():
    print array
    print array.sum()

if __name__ == "__main__":
    test_run()

Here is sum.

[[1 2 3]
 [2 3 4]]
15

Sum of each column and row

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4)])
    print array
    print array.sum(0) # sum of columns
    print array.sum(1) # sum of rows

if __name__ == "__main__":
    test_run()

Here are the sums.

[3 5 7]
[6 9]

Finding max, min and mean in an array

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4)])
    print array

    print array.max(0) # max of first column
    print array.min(0) # min of first column
    print array.mean() # mean of all elements

if __name__ == "__main__":
    test_run()

Her are the values.

[[1 2 3]
 [2 3 4]]
[2 3 4]
[1 2 3]
2.5

There are more functions available, check this page.

Find index of max value

import numpy as np

def get_max_index(a):
    return np.argmax(a)


def test_run():
    a = np.array([9, 6, 2, 3, 12, 14, 7, 10], dtype=np.int32)  # 32-bit integer array
    print "Array:", a

    # Find the maximum and its index in array
    print "Maximum value:", a.max()
    print "Index of max.:", get_max_index(a)

if __name__ == "__main__":
    test_run()

Here is the output.

Array: [ 9  6  2  3 12 14  7 10]
Maximum value: 14
Index of max.: 5

Glue arrays together

There are many ways to append or connect two arrays. Depends what we really need.

Append one array to another array.

In [1]: import numpy as np

In [2]: a = np.array([[1, 2, 3], [4, 5, 6]])

In [3]: b = np.array([[9, 8, 7], [6, 5, 4]])

In [4]: np.concatenate((a, b))
Out[4]: 
array([[1, 2, 3],
       [4, 5, 6],
       [9, 8, 7],
       [6, 5, 4]])

Or we can do the same thing using vstack function.

In [1]: a = np.array([1, 2, 3])

In [2]: b = np.array([4, 5, 6])

In [3]: np.vstack((a, b))
Out[3]: 
array([[1, 2, 3],
       [4, 5, 6]])

Or using hstack to do it horizontally.

a = np.array((1,2,3))
b = np.array((2,3,4))
np.hstack((a,b))

a = np.array([[1],[2],[3]])
b = np.array([[2],[3],[4]])
np.hstack((a,b))

Add another column into an array.

my_data = np.random.random((210,8))
new_col = my_data.sum(1)[...,None]
new_col.shape
#(210,1)
all_data = np.append(my_data, new_col, 1)
all_data.shape
#(210,9)

# or
>>> a = np.array([[1,2,3],[2,3,4]])
>>> a
array([[1, 2, 3],
       [2, 3, 4]])
>>> z = np.zeros((2,1))
>>> z
array([[ 0.],
       [ 0.]])
>>> np.concatenate((a, z), axis=1)
array([[ 1.,  2.,  3.,  0.],
       [ 2.,  3.,  4.,  0.]])

Slicing array

If you want to slice data, get some specific values of array, you can access it via indices.

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4), (4, 5, 6)])
    print array
    print array[:, 0:3:2]

if __name__ == "__main__":
    test_run()

First argument ":" makes sure all rows are included. Second argument if more complex. 0 and 3 are saying what values to take. 2 is saying what will be skipped. Here is the output.

[[1 2 3]
 [2 3 4]
 [4 5 6]]
[[1 3]
 [2 4]
 [4 6]]

Assigning values in array

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4), (4, 5, 6)])
    print array
    array[0][0] = 100
    print  array

if __name__ == "__main__":
    test_run()

Here is the output.

[[1 2 3]
 [2 3 4]
 [4 5 6]]
[[100   2   3]
 [  2   3   4]
 [  4   5   6]]

Or you can assign value to whole row or column.

import numpy as np

def test_run():
    array = np.array([(1, 2, 3), (2, 3, 4), (4, 5, 6)])
    print array
    array[:, 0] = 100
    print  array

if __name__ == "__main__":
    test_run()

Here is the output.

[[1 2 3]
 [2 3 4]
 [4 5 6]]
[[100   2   3]
 [100   3   4]
 [100   5   6]]

Or you can set list of values to a row.

array[:, 0] = [0, 0,0]

Indexing

import numpy as np

def test_run():
    array = np.random.rand(5)

    indices = np.array([1, 1, 2, 3])

    print array
    print array[indices]

if __name__ == "__main__":
    test_run()

Here is the output.

[ 0.22607416  0.94175441  0.45115583  0.3740176   0.53811653]
[ 0.94175441  0.94175441  0.45115583  0.3740176 ]

Filtering data using conditions

Also called masking.

import numpy as np

def test_run():
    array = np.array([(20, 25, 10, 5), (10, 25, 10, 5)])

    print array

    mean = array.mean()
    print mean

    print array[array < mean]

if __name__ == "__main__":
    test_run()

Output.

[[20 25 10  5]
 [10 25 10  5]]
13.75
[10  5 10 10  5]

Multiplication

import numpy as np

def test_run():
    array = np.array([(20, 25, 10, 5), (10, 25, 10, 5)])

    print array
    print array * 2

if __name__ == "__main__":
    test_run()

Output.

[[20 25 10  5]
 [10 25 10  5]]
[[40 50 20 10]
 [20 50 20 10]]

Sum arrays

import numpy as np

def test_run():
    array = np.array([(20, 25, 10, 5), (10, 25, 10, 5)])

    print array
    print array + array

if __name__ == "__main__":
    test_run()

Output.

[[20 25 10  5]
 [10 25 10  5]]
[[40 50 20 10]
 [20 50 20 10]]