Reshaping arrays
- Machine Learning course
- Beginner
- Lesson 11/112
- 5 min
The reshape() method changes the array's shape while keeping the total element count.
reshape
reshape gives an array a new shape without changing the number of elements. The product of the new shape must equal the old size:
a = np.arange(6) # [0 1 2 3 4 5]
print(a.reshape(2, 3)) # [[0 1 2] [3 4 5]]
print(a.reshape(3, -1)) # -1: work out the columns -> (3, 2)
# a.reshape(4, 2) ValueError: 6 elements don't fit into 8The elements are placed row by row (C order).
flatten and transpose
flatten() reduces the array to one dimension; .T turns rows into columns:
m = np.array([[1, 2, 3],
[4, 5, 6]])
print(m.flatten()) # [1 2 3 4 5 6]
print(m.T.shape) # (3, 2)
print(m.T) # [[1 4] [2 5] [3 6]]With image data, turning 28×28 pixels into a vector of length 784 before giving it to a neural network is a common step.
Practice Reshape
Turn the array np.arange(1, 13) into a matrix with 3 rows (use -1). Print the matrix, then the shape of its transpose.
Starter code
import numpy as np
a = np.arange(1, 13)
# buraya yazSample solution
import numpy as np
a = np.arange(1, 13)
m = a.reshape(3, -1)
print(m)
print(m.T.shape)Expected output
[[ 1 2 3 4] [ 5 6 7 8] [ 9 10 11 12]] (4, 3)
Questions and answers
Review the lesson questions together with their answers. To test yourself, solve the lesson interactively.
1 Which is invalid for an array of 12 elements?
- reshape(2, -1)
- reshape(2, 2, 3)
- reshape(5, -1) Correct answer
- reshape(3, 4)
Explanation: 12 isn't divisible by 5, so -1 can't be worked out and a ValueError occurs.
2 Reduce the array to one dimension.
v = m.____()v = m.flatten()Explanation: flatten always returns a new one-dimensional copy.
3 Match the operations with the resulting shapes (a = np.arange(12)).
| Term | Meaning |
|---|---|
a.reshape(3, 4) |
(3, 4) |
a.reshape(3, 4).T |
(4, 3) |
a.reshape(2, -1) |
(2, 6) |
a.reshape(2, 6).flatten() |
(12,) |