IKH

Menu

Chepter-13

Searching elements of ndarray

  • We can search elements of ndarray by using where() function.

where(condition, [x, y])

  • If we specify only the condition, then it will return the indices of the elements which satisfies the condition.
  • If we provide condition,x,y then The elements which satisfies the condition will be replaced with x and remaining elements will be replaced with y
  • where() function does not return the elements. It returns only the indices
  • it will act as replacement operator also.
  • The functionality is just similar to ternary operator when it acts as replacement
    operator

Example

Python
In [340]: 
import numpy as np 
help(np.where)

Output

PowerShell
Help on function where in module numpy: 
 
where(...) 
    where(condition, [x, y]) 
     
    Return elements chosen from `x` or `y` depending on `condition`.

where() function

Example

Python
In [341]: 
# Find indexes where the value is 7 from 1-D array 
a = np.array([3,5,7,6,7,9,4,6,10,15]) 
b = np.where(a==7) 
b # element 7 is available at 2 and 4 indices

Output

PowerShell
Out[341]: 
(array([2, 4], dtype=int64),)

Example

Python
In [342]: 
# Find indices where odd numbers present in the given 1-D array? 
a = np.array([3,5,7,6,7,9,4,6,10,15]) 
b = np.where(a%2!=0) 
b

Output

PowerShell
Out[342]: 
(array([0, 1, 2, 4, 5, 9], dtype=int64),)

Finding the elements directly

We can get the elements directly in 2 ways

  • using where() function
  • using condition based selection
where() function

Example

Python
In [343]: 
# to get the odd numbers 
a = np.array([3,5,7,6,7,9,4,6,10,15]) 
indices = np.where(a%2!=0) 
a[indices]

Output

PowerShell
Out[343]: 
array([ 3,  5,  7,  7,  9, 15])

conditional based selection

Example

Python
In [344]: 
# to get the odd numbers 
a = np.array([3,5,7,6,7,9,4,6,10,15]) 
a[a%2!=0]

Output

PowerShell
Out[344]: 
array([ 3,  5,  7,  7,  9, 15])

Example

Python
In [345]: 
# where(condition,[x,y]) 
# if condition satisfied that element will be replaced from x and  
# if the condition fails that element will be replaced from y. 
# Replace every even number with 8888 and every odd number with 7777? 
a = np.array([3,5,7,6,7,9,4,6,10,15]) 
b = np.where( a%2 == 0, 8888, 7777) 
b

Output

PowerShell
Out[345]: 
array([7777, 7777, 7777, 8888, 7777, 7777, 8888, 8888, 8888, 7777])

Example

Python
In [346]: 
# Find indexes where odd numbers present in the given 1-D array and replace with 
element 9999. 
a = np.array([3,5,7,6,7,9,4,6,10,15]) 
b = np.where( a%2 != 0, 9999, a) 
b

Output

PowerShell
Out[346]: 
array([9999, 9999, 9999,    6, 9999, 9999,    4,    6,   10, 9999])

We can use where() function for any n-dimensional array

2-D arrays

  • It will return the 2 arrays.
  • First array is row indices
  • Second array is column indices

Example

Python
In [347]: 
# to find the indices of the elements where elements are divisible by 5 
a = np.arange(12).reshape(4,3) 
np.where(a%5==0)

Output

PowerShell
Out[347]: 
(array([0, 1, 3], dtype=int64), array([0, 2, 1], dtype=int64))
  • The first array array([0, 1, 3] represents the row indices
  • The second array array([0, 2, 1] represents the column indices
  • The required elements present at (0,0),(1,2) and (3,1) index places.

Example

Python
In [348]: 
# we can perform replacement on 2-D arrays 
a = np.arange(12).reshape(4,3) 
np.where(a%5==0,9999,a)

Output

PowerShell
Out[348]: 
array([[9999,    1,    2], 
       [   3,    4, 9999], 
       [   6,    7,    8], 
       [   9, 9999,   11]])

searchsorted() function

  • Internally this function will use Binary Search algorithm. Hence we can call this
  • function only for sorted arrays.
  • If the array is not sorted then we will get abnormal results.
  • Complexicity of the Binary search algorithm is O(log n)
  • It will return insertion point(i.e., index) of the given element

Example

Python
In [349]: 
import numpy as np 
help(np.searchsorted)

Output

PowerShell
Help on function searchsorted in module numpy: 
 
searchsorted(a, v, side='left', sorter=None) 
    Find indices where elements should be inserted to maintain order.

Example

Python
In [350]: 
# to find the insertion point of 6 from left 
a = np.arange(0,31,5) 
np.searchsorted(a,6)

Output

PowerShell
Out[350]: 
2

Note

  • Bydefault it will always search from left hand side to identify insertion point.
  • If we want to search from right hand side we should use side=’right’

Example

Python
In [351]: 
# to find the insertion point of 6 from right 
a = np.arange(0,31,5) 
print(f"Array a :\n {a}") 
np.searchsorted(a,6,side='right')

Output

PowerShell
Array a : 
 [ 0  5 10 15 20 25 30] 
Out[351]: 
2

Example

Python
In [352]: 
# to find the insetion point from left and right 
a = np.array([3,5,7,6,7,9,4,10,15,6]) 
# first sort the elements 
a = np.sort(a) 
 
# insertion point from left(default) 
left = np.searchsorted(a,6) 
# insertion point from right 
right = np.searchsorted(a,6,side='right') 
 
print(f"The original array : {a}") 
print(f"Insertion point for 6 from left : {left}") 
print(f"Insertion point for 6 from right : {right}")

Output

PowerShell
The original array : [ 3  4  5  6  6  7  7  9 10 15] 
Insertion point for 6 from left : 3 
Insertion point for 6 from right : 5

Summary:

  • sort() ==> To sort given array
  • where() ==> To perform search and replace operation
  • searchsorted() ==> To identify insertion point in the given sorted array

Report an error

Prev
Next