Introduction
In knowledge manipulation and scientific computing, NumPy stands as one of many most-used libraries because it offers fairly a number of functionalities. One such operation, “flattening,” helps to rework multi-dimensional arrays right into a one-dimensional sequence. Whereas flattening a complete array is fairly easy, there are occasions once you may wish to selectively flatten particular dimensions to go well with the necessities of your knowledge pipeline or algorithm. On this Byte, we’ll see varied methods to realize this extra nuanced type of flattening.
NumPy Arrays
NumPy, brief for Numerical Python, is a library in Python that gives assist for giant, multi-dimensional arrays and matrices, together with a group of mathematical capabilities to function on these arrays. NumPy arrays are a key ingredient in scientific computing with Python. They’re extra environment friendly and sooner in comparison with Python’s built-in checklist knowledge sort, particularly in the case of mathematical operations.
This code reveals what a NumPy array can appear like:
import numpy as np
# Making a 2D NumPy array
array_2D = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print(array_2D)
Output:
[[1 2 3]
[4 5 6]
[7 8 9]]
Why Flatten Some Dimensions of a NumPy Array?
Flattening an array means changing a multidimensional array right into a 1D array. However why would you wish to flatten just a few dimensions of a NumPy array?
Effectively, there are numerous situations the place you may want to do that. For instance, in machine studying, usually we have to flatten our enter knowledge earlier than feeding it right into a mannequin. It’s because many machine studying algorithms anticipate enter knowledge in a selected format, often as a 1D array.
However generally, you won’t wish to flatten your entire array. As a substitute, you may wish to flatten particular dimensions of the array whereas maintaining the opposite dimensions intact. This may be helpful in situations the place you wish to preserve some stage of the unique construction of the info.
How you can Flatten a NumPy Array
Flattening a NumPy array is pretty straightforward to do. You need to use the flatten() technique offered by NumPy to flatten an array:
import numpy as np
# Making a 2D NumPy array
array_2D = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
# Flattening the 2D array
flattened_array = array_2D.flatten()
print(flattened_array)
Output:
[1 2 3 4 5 6 7 8 9]
As you possibly can see, the flatten() technique has remodeled our 2D array right into a 1D array.
However what if we wish to flatten solely a selected dimension of the array and never your entire array? We’ll discover this within the subsequent sections.
Flattening Particular Dimensions of a NumPy Array
Flattening a NumPy array is sort of easy. However, what if it’s good to flatten solely particular dimensions of an array? That is the place the reshape perform comes into play.
To illustrate we now have a 3D array and we wish to flatten the final two dimensions, maintaining the primary dimension as it’s. The reshape perform can be utilized to realize this. This is a easy instance:
import numpy as np
# Create a 3D array
array_3d = np.array([[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]])
# Reshape the array
flattened_array = array_3d.reshape(array_3d.form[0], -1)
print(flattened_array)
Output:
[[ 1 2 3 4 5 6]
[ 7 8 9 10 11 12]]
Within the above code, the -1 within the reshape perform signifies that the dimensions of that dimension is to be calculated mechanically. That is primarily based on the dimensions of the array and the dimensions of the opposite dimensions.
Be aware: The reshape perform doesn’t modify the unique array. As a substitute, it returns a brand new array that has the required form.
Comparable Options and Use-Instances
Flattening particular dimensions of a NumPy array is not the one approach to manipulate your knowledge. There are different related options you may discover helpful. For instance, the ravel perform may also be used to flatten an array. Nonetheless, in contrast to reshape, ravel at all times returns a flattened array.
Moreover, you should utilize the transpose perform to alter the order of the array dimensions. This may be helpful in circumstances the place it’s good to rearrange your knowledge for particular operations or visualizations.
These methods might be significantly helpful in knowledge preprocessing for machine studying. As an illustration, you may must flatten the enter knowledge for a neural community. Or, you may must transpose your knowledge to make sure that it is within the right format for a specific library or mathematical perform.
Conclusion
On this Byte, we have explored how you can flatten particular dimensions of a NumPy array utilizing the reshape perform. We have additionally checked out related options similar to ravel and transpose and mentioned some use-cases the place these methods might be significantly helpful.
Whereas these methods are highly effective instruments for knowledge manipulation, they’re simply the tip of the iceberg in the case of what you are able to do with NumPy. So I might recommend taking a deeper have a look at the NumPy documentation and see what different fascinating options you possibly can uncover.
