How to iterate over Columns in Numpy

NumPy provides several methods to iterate over columns in a 2D array. The most common approaches include using nditer() with transpose, array transpose directly, apply_along_axis(), and manual iteration with indexing.

Syntax

Here are the key functions used for column iteration ?

np.nditer(array.T)          # Iterator with transpose
array.T                     # Array transpose  
np.apply_along_axis()       # Apply function along axis
array.shape[1]              # Get number of columns

Using nditer() with Transpose

The nditer() function creates an iterator object that can traverse array elements. Combined with transpose, it iterates through each column ?

import numpy as np

# Create a sample 2D array
arr = np.array([[10, 20, 30], [40, 50, 60], [70, 80, 90], [110, 120, 130]])
print("Original array:")
print(arr)
print("\nIteration over columns using nditer:")

# Iterate over columns using nditer with transpose
for col in np.nditer(arr.T):
    print(col)

Original array:
[[ 10  20  30]
 [ 40  50  60]
 [ 70  80  90]
 [110 120 130]]

Iteration over columns using nditer:
10
40
70
110
20
50
80
120
30
60
90
130

Using Array Transpose

The transpose operation converts rows to columns, allowing direct iteration over columns as complete arrays ?

import numpy as np

# Create a 3x3 matrix
arr = np.array([[1, 2, 3],
                [4, 5, 6], 
                [7, 8, 9]])

print("Iterating over columns using transpose:")
# Transpose and iterate over each column
for col in arr.T:
    print(col)

Iterating over columns using transpose:
[1 4 7]
[2 5 8]
[3 6 9]

Using apply_along_axis() Function

The apply_along_axis() function applies a given function to 1D slices along a specified axis ?

import numpy as np

# Create a 3x3 array
arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

# Function to process each column
def process_column(col):
    print(f"Column: {col}")
    return col

print("Using apply_along_axis to iterate columns:")
# Apply function along axis 0 (columns)
np.apply_along_axis(process_column, 0, arr)

Using apply_along_axis to iterate columns:
Column: [1 4 7]
Column: [2 5 8]
Column: [3 6 9]

Using Manual Indexing

You can manually iterate through columns using array indexing and the shape property ?

import numpy as np

# Create a 3x3 array
arr = np.array([[1, 5, 9], [13, 14, 21], [25, 29, 33]])

# Get the number of columns
num_cols = arr.shape[1]

print("Manual iteration over columns:")
# Iterate through each column index
for col_index in range(num_cols):
    col = arr[:, col_index]  # Extract column
    print(f"Column {col_index}: {col}")

Manual iteration over columns:
Column 0: [ 1 13 25]
Column 1: [ 5 14 29]
Column 2: [ 9 21 33]

Comparison of Methods

Conclusion

Use arr.T for simple column iteration, apply_along_axis() for applying functions to columns, and manual indexing when you need both column data and indices. Choose the method based on whether you need individual elements or complete columns.