8  Slicing Data

Slicing is the process of selecting a subset of an array.

8.1 Slicing in 1-D

Given a 1-D NumPy array named vec, the syntax to select a range of indices is: \[\text{vec} [\text{start} : \text{stop} : \text{step}]\] where \(\text{start}\) is the starting index, \(\text{stop}\) is the excluded stopping index, and \(\text{step}\) is an incriment. Consider the array:

vec = np.array([10, 20, 30, 40, 50, 60, 70, 80])

which has the structure:

The following examples are slices of vec:

Slice	Result	Description
vec[1:4]	[20, 30, 40]	Elements from index 1 up to 4 (excluded)
vec[:3]	[10, 20, 30]	Elements up to index 3 (excluded)
vec[3:]	[40, 50, 60, 70, 80]	Elements from index 3 to the end
vec[1:5:3]	[20, 50]	From index 1 to 5 (excluded) skipping by 3
vec[2::2]	[30, 50, 70]	Elements from 2 to the end, skipping by 2

Specifying Indices

We are not constrained to selecting only sequenced elements from an array. We can select arbitrary elements by passing a list of which indices we want to select:

import numpy as np
vec = np.array([10, 20, 30, 40, 50, 60, 70, 80])
vec[ [1,5,6] ]

array([20, 60, 70])

8.2 Negative Indices

Negative indices count backwards, so a[-1] refers to the last element in a and a[-2] references the second to last element in a. Stepping by \(-1\) means moving backwards through an array, so that we can reverse an array with:

import numpy as np 

vec = np.array([10, 20, 30, 40, 50, 60, 70, 80])

print( vec[::-1] )

[80 70 60 50 40 30 20 10]

8.3 Slicing in 2-D

Slicing a 2-D array requires specifying rows and columns. Consider an array arr:

arr = np.array([[10,  20,  30,  40],
                [50,  60,  70,  80],
                [90, 100, 110, 120]])

which has the structure:

We can select the middle row by specifying row 1 and using : to reprsent “all” columns:

arr[1,:]

array([50, 60, 70, 80])

We can select the last column by specifying any row : and column 3:

arr[:,3]

array([ 40,  80, 120])

or equivilently choosing the last column:

arr[:,-1]

array([ 40,  80, 120])

The value of \(70\) is located at row 1 column 2, which we can select with:

arr[1,2]

70

Notice that this returns the value \(70\) and not a subarray containing the value 70. Indexing a single element within an array references the value at that position.

8.4 Updating Values

8.4.1 Update an element

Updating the value of a single element

arr = np.array([[10,  20,  30,  40],
                [50,  60,  70,  80],
                [90, 100, 110, 120]])

arr[1,2] = -1
print(arr)

[[ 10  20  30  40]
 [ 50  60  -1  80]
 [ 90 100 110 120]]

8.4.2 Update Rows or Columns

We can update any row, column or the entire array with the same operations we used in Chapter 4. Given an array arr,

arr = np.array([[10,  20,  30,  40],
                [50,  60,  70,  80],
                [90, 100, 110, 120]])
arr[2] = arr[2] * 3
print(arr)                

[[ 10  20  30  40]
 [ 50  60  70  80]
 [270 300 330 360]]

8.5 Graphing Data from Slices

Data will commonly be in array where each row corresponds to a sample and each column is a feature.

data = []

Exercises

1. Use slicing to set all of the non-zero elements of M to \(1\):

   M =  np.array([[ 0, 0, 0, 0, 0, 0, 0, 0],
                  [ 0, 0, 1, 7, 0, 0, 0, 0],
                  [ 0, 0, 4, 3, 0, 0, 0, 0],
                  [ 0, 0, 9, 2, 0, 0, 0, 0],
                  [ 0, 0, 0, 0, 0, 0, 0, 0]])

2. Given a 1-D array named TwoHundo containing two hundred random digits, write the one line of code that would set every other element of TwoHundo equal to \(0\), beginning with the second element. Create such an array and demonstrate your working one-liner.

3. Row operations?

4. Complete the penguin classifier in this Colab document: https://colab.research.google.com/drive/1Cb9iHNO1fA4BPO-puaTb7B64V4mf9veC?usp=sharing