How to Flatten a Matrix using numpy in Python?

In this article, we will show you how to flatten a matrix using the NumPy library in python.

numpy.ndarray.flatten() Function

The numpy module includes a function called numpy.ndarray.flatten() that returns a one-dimensional copy of the array rather than a two-dimensional or multi-dimensional array.

In simple words, we can say that it flattens a matrix to 1-Dimension.

Syntax

ndarray.flatten(order='C')

Parameters

order ? 'C', 'F', 'A', 'K' (optional)

• When we set the order parameter to 'C', the array is flattened in row-major order.

• When the 'F' is set, the array is flattened in column-major order.

• When the order parameter is set to 'A', the array is flattened in column-major order only if 'a' is Fortran contiguous in memory. The 'K' order flattens the array in the same order that the elements appeared in memory. This parameter is set to 'C' by default.

Return Value ? Returns a flattened 1-D matrix

Using flatten() on 2D Matrix

The following program flattens the given input 2-Dimensional matrix to a 1-Dimensional matrix using the flatten() function ?

# importing numpy module with an alias name
import numpy as np

# creating a 2-Dimensional(2x2) numpy matrix
inputMatrix = np.array([[3, 5], [4, 8]])

# printing the input 2D matrix
print("The input numpy matrix:")
print(inputMatrix)

# flattening the 2D matrix to one-dimensional matrix
flattenMatrix = inputMatrix.flatten()

# printing the resultant flattened matrix
print("Resultant flattened matrix:")
print(flattenMatrix)

The input numpy matrix:
[[3 5]
 [4 8]]
Resultant flattened matrix:
[3 5 4 8]

Using reshape() Function

The following program flattens the given input 4-Dimensional matrix to a 1-Dimensional matrix using reshape() function ?

# importing numpy module with an alias name
import numpy as np

# creating a 4-Dimensional(4x4) numpy matrix
inputMatrix = np.array([[1, 2, 3, 97],
                       [4, 5, 6, 98],
                       [7, 8, 9, 99],
                       [10, 11, 12, 100]])

# printing the input 4D matrix
print("The input numpy matrix:")
print(inputMatrix)

# reshaping the array and flattening the 4D matrix to a one-dimensional matrix
flattenMatrix = np.reshape(inputMatrix, -1)

# printing the resultant flattened matrix
print("Resultant flattened matrix:")
print(flattenMatrix)

The input numpy matrix:
[[  1   2   3  97]
 [  4   5   6  98]
 [  7   8   9  99]
 [ 10  11  12 100]]
Resultant flattened matrix:
[  1   2   3  97   4   5   6  98   7   8   9  99  10  11  12 100]

Using flatten() on np.matrix Type

The following program flattens the given input matrix created with np.matrix() using the flatten() function ?

# importing NumPy module with an alias name
import numpy as np

# creating a NumPy matrix (4x4 matrix) using matrix() method
inputMatrix = np.matrix('11, 1, 8, 2; 11, 3, 9, 1; 1, 2, 3, 4; 9, 8, 7, 6')

# printing the input 4D matrix
print("The input numpy matrix:")
print(inputMatrix)

# flattening the 4D matrix to one-dimensional matrix
flattenMatrix = inputMatrix.flatten()

# printing the resultant flattened matrix
print("Resultant flattened matrix:")
print(flattenMatrix)

The input numpy matrix:
[[11  1  8  2]
 [11  3  9  1]
 [ 1  2  3  4]
 [ 9  8  7  6]]
Resultant flattened matrix:
[[11  1  8  2 11  3  9  1  1  2  3  4  9  8  7  6]]

Comparison of Methods

Conclusion

Use flatten() for creating a copy of flattened data, reshape(-1) for memory-efficient flattening, or ravel() for the fastest flattening operation. All methods convert multi-dimensional arrays into 1D arrays.