Numpy - Dimensional Functions

Numpy Dimension

Dimension

---

---

Reshaping

Reshaping an array changes the shape of the data without altering the data itself.

Reshaping arrays requires that the number of data elements remains constant

Either syntax below can be used:

result = np.reshape(a, newshape)

result = a.reshape(newshape)

Gives a new shape to a NumPy array a without changing its data.

Converting 1D Arrays to 2D Arrays

new_array = array.reshape((dimension),(elements))

if, dimension = 1, elements = 5 and based array is [[1,2,3,4,5,6]]
the output array will be [[1,2,3,4,5,6]]
and if we do below..

new_array = array.reshape(5,1)

the output array will be
[[1]
[2]
[3]
[4]
[5]
[6]

now, we will re shape the array to 2x3

new_array = array.reshape(2,3)

the output will be showing like, [1,2,3]
[4,5,6]

array[1] = [4,5,6]

Reshaping Multi-Dimension Arrays

Let’s said,
There is a simple array for example.
Array:
[[[10 11]
[10 12]
[10 13]
[10 14]
[10 15]]

[[20 21]
[20 22]
[20 23]
[20 24]
[20 25]]

[[30 31]
[30 32]
[30 33]
[30 34]
[30 35]]]

we’ll gonna change the dimension to 3,10

new_array = array.reshape(3,10)

the output array will be shoing like,
Array:
[[10 11 10 12 10 13 10 14 10 15]
[20 21 20 22 20 23 20 24 20 25]
[30 31 30 32 30 33 30 34 30 35]]

Adding an Axis or Dimension

Sometimes, we are facing the situation add axis or dimension for data processing

Adding a dimension to an exsting Numpy array is always possible since it does not changed the number of elements in the array.

Let’s see what happens when we add the dimension.

Here is based array.
Array:
[[1 2 3]
[4 5 6]
[7 8 9]]

Array dimension: 2

new_array = array.reshape(1,3,3)

Output Array:
[[[1 2 3]
[4 5 6]
[7 8 9]]]

Array Dimension: 3

Expand Dimensions

In the previous section, we used reshape to add a new axis. We can use np.expand_dims or np.newaxis to do the same thing.

np.expand_dims(a, axis)

Expand the shape of an array. Insert a new axis that will appear at the axis position in the expanded array shape.

# Create a 1D array.
print('Create a 1D array:')
c = np.arange(1, 10, dtype=int)
array_info(c)

# Reshape to a 3x3 2D array.
c = c.reshape(3,3)
array_info(c)

# Expand dimensions using: np.expand_dims
print('Using np.expand_dims:')
c_expand = np.expand_dims(c, axis=0)
array_info(c_expand)

# Expand dimensions using: np.newaxis
print('Using np.newaxis:')
c_newaxis = c[np.newaxis, :, : ]
array_info(c_newaxis)

# You can also use `None`, but np.newaxis is prefered since it is more explicit.
# print('Using None:')
# c_none = c[None, :, : ]
# array_info(c_none)

Output:
Create a 1D array:
Array:
[1 2 3 4 5 6 7 8 9]
Data type: int32
Array shape: (9,)
Array Dim: 1

Array:
[[1 2 3]
[4 5 6]
[7 8 9]]
Data type: int32
Array shape: (3, 3)
Array Dim: 2

Using np.expand_dims:
Array:
[[[1 2 3]
[4 5 6]
[7 8 9]]]
Data type: int32
Array shape: (1, 3, 3)
Array Dim: 3

Using np.newaxis:
Array:
[[[1 2 3]
[4 5 6]
[7 8 9]]]
Data type: int32
Array shape: (1, 3, 3)
Array Dim: 3

Compare btwn reshap and expand_dims

Feature	np.reshape()	np.expand_dims()
Purpose	Changes the entire shape of the array	Adds a new dimension of size 1 at a specified axis
Input Format	Desired shape as a tuple or list of integers	Axis where the new dimension is to be inserted
Automatic Dimension Inference	❌ Requires all dimensions specified explicitly	✅ Adds a singleton dimension without affecting existing data
Example	(2, 3) → (3, 2)	(3,) → (1, 3) or (3, 1)
Common Use Case	Reshaping data for models	Adding batch or channel dimensions

Squeeze

np.squeeze(a, axis=None)
Remove axes of length one from a.

output_array = array.squeeze(e, axis=0)

a = Target array
axis = index that you want to delete

Reshape Revisit

reshape_1 = array.reshape(-1,2)
reshape_2 = array.reshape(2,-1)

Based array:
[0 1 2 3 4 5 6 7 8 9]

Reshape 1 array: [[0 1]
[2 3]
[4 5]
[6 7]
[8 9]]

Reshape 2 array:
[[0 1 2 3 4]
[5 6 7 8 9]]

Flattening Arrays

new_array = array.reshape(1,3,16)

Based array:
[[[1. 1. 1. 1.]
[1. 1. 1. 1.]
[1. 1. 1. 1.]
[1. 1. 1. 1.]] \

[[2. 2. 2. 2.]
[2. 2. 2. 2.]
[2. 2. 2. 2.]
[2. 2. 2. 2.]] \

[[3. 3. 3. 3.]
[3. 3. 3. 3.]
[3. 3. 3. 3.]
[3. 3. 3. 3.]]]

Output Array:
[[[1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1.]
[2. 2. 2. 2. 2. 2. 2. 2. 2. 2. 2. 2. 2. 2. 2. 2.]
[3. 3. 3. 3. 3. 3. 3. 3. 3. 3. 3. 3. 3. 3. 3. 3.]]]