Numpy arrays look and act very much like other Python sequences…
…but they're technically not sequences
They're iterable. They're containers. They're sized. But they're not sequences
A sequence is a collection in which you can access an element using an integer index
You can do this with NumPy arrays:
>>> some_array = np.array([3, 6, 9])
>>> some_array[1]
6
But you can also use more advanced indexing, such as:
>>> some_array = np.array([[1, 2, 3], [4, 5, 6]])
>>> some_array
array([[1, 2, 3],
[4, 5, 6]])
>>> some_array[1, 2]
6
You use two indices to index the rows and columns
Here's more in indexing in NumPy in this 2-day series:
twitter.com/s_gruppetta_ct/status/1624034209936609281
However, there's another important difference you should be aware of when moving from using lists to NumPy arrays
Here's an example to show the difference in behaviour:
Let's start with a list:
>>> numbers = [2, 4, 6, 8, 10]
…and get a slice from that list:
>>> numbers_subset = numbers[2:5]
>>> numbers_subset
[6, 8, 10]
Change a value from the slice:
>>> numbers_subset[-1] = 999
>>> numbers_subset
[6, 8, 999]
As you'd expect, the last value of `numbers_subset` has changed, but…
The original list has _not_ changed:
>>> numbers
[2, 4, 6, 8, 10]
When you create a slice of a list or any sequence, you're creating a _copy_ of the data
Let's replicate this with NumPy arrays:
>>> import numpy as np
>>> numbers = np.array([2, 4, 6, 8, 10])
>>> numbers_subset = numbers[2:5]
>>> numbers_subset
array([ 6, 8, 10])
You create an array and a slice of that array. So far, nothing _looks_ different
You can change the value of one of the elements from the slice:
>>> numbers_subset[-1] = 999
>>> numbers_subset
array([ 6, 8, 999])
This should be what you were expecting, too. But…
If you look at `number`, which is the original array, you'll note different behaviour from when you used lists:
>>> numbers
array([ 2, 4, 6, 8, 999])
The original array has also changed when you changed a value in the slice
You made the change in `numbers_subset` but the value also changed in `numbers`
Why?
When you create a slice on a NumPy array, you create a "view" and not a "copy" of the array – you're still using the same data in the original array
You can read more on NumPy views here (in this rather old thread…):
twitter.com/s_gruppetta_ct/status/1522301373592653824
And if you want to make sure you understand the difference between iterable, collection, sequence, and other terms we use for data structures, you can see this series:
twitter.com/s_gruppetta_ct/status/1628341581538549760
Stephen Gruppetta
@s_gruppetta_ct
Constantly looking for innovative ways to talk and write about Python • Mentoring learners • Writing about Python • Writing about technical writing