Aggregation along specific axes in NumPy - Time & Space Complexity

Time Complexity: Aggregation along specific axes

O(n * m)

Understanding Time Complexity

When we use numpy to add or average numbers along rows or columns, it takes some time depending on the data size.

We want to know how this time grows as the data gets bigger.

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

import numpy as np

arr = np.random.rand(1000, 500)
sum_rows = np.sum(arr, axis=1)
sum_cols = np.sum(arr, axis=0)
mean_rows = np.mean(arr, axis=1)
mean_cols = np.mean(arr, axis=0)

This code creates a 2D array and calculates sums and averages along rows and columns.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: Adding or averaging all elements along one axis (rows or columns).
• How many times: For each row or column, numpy visits every element once.

How Execution Grows With Input

When the array size grows, the number of elements to add or average grows too.

Input Size (rows x cols)	Approx. Operations
10 x 10	100 additions
100 x 100	10,000 additions
1000 x 1000	1,000,000 additions

Pattern observation: The operations grow roughly with the total number of elements in the array.

Final Time Complexity

Time Complexity: O(n * m)

This means the time grows proportionally to the total number of elements in the array.

Common Mistake

[X] Wrong: "Summing along one axis only takes time proportional to the number of rows or columns, not the whole array."

[OK] Correct: Even when summing along one axis, numpy must visit every element in that axis, so it still processes all elements in the array.

Interview Connect

Understanding how numpy processes data helps you explain performance in real tasks like data analysis or machine learning.

Self-Check

"What if we used numpy's sum on a 3D array along one axis? How would the time complexity change?"

Practice

1. What does setting axis=0 do when aggregating a 2D NumPy array?

easy

A. Aggregates values down each column (per column)

B. Aggregates values across each row (per row)

C. Flattens the array before aggregation

D. Aggregates only the diagonal elements

Solution

1. Step 1: Understand axis parameter meaning

   In NumPy, axis=0 means aggregation happens down the rows, combining values in each column.

2. Step 2: Apply to 2D array

   Aggregating with axis=0 summarizes each column separately.

3. Final Answer:

   Aggregates values down each column (per column) -> Option A

4. Quick Check:

   axis=0 means down rows = per column aggregation [OK]

Hint: axis=0 aggregates columns, axis=1 aggregates rows [OK]

Common Mistakes:

• Confusing axis=0 with axis=1
• Thinking axis=0 aggregates rows instead of columns
• Assuming axis=0 flattens the array

2. Which of the following is the correct syntax to sum a 2D NumPy array arr along rows (per row)?

easy

A. np.sum(arr, axis=1)

B. np.sum(arr, axis=0)

C. arr.sum(axis=2)

D. np.sum(arr)

Solution

1. Step 1: Identify axis for row-wise sum

   Summing across columns (per row) requires axis=1.

2. Step 2: Check syntax correctness

   np.sum(arr, axis=1) correctly sums each row's elements.

3. Final Answer:

   np.sum(arr, axis=1) -> Option A

4. Quick Check:

   Sum per row uses axis=1 [OK]

Hint: Use axis=1 to sum across columns (per row) [OK]

Common Mistakes:

• Using axis=0 to sum rows
• Using invalid axis like 2 for 2D array
• Forgetting to specify axis

3. Given the array:

arr = np.array([[1, 2, 3], [4, 5, 6]])

What is the output of np.sum(arr, axis=0)?

medium

A. [3 7 11]

B. [5 7 9]

C. [6 15]

D. [1 2 3 4 5 6]

Solution

1. Step 1: Understand axis=0 sum

   Sum down rows means add elements in each column: (1+4), (2+5), (3+6).

2. Step 2: Calculate sums

   Calculations: 1+4=5, 2+5=7, 3+6=9, so result is [5 7 9].

3. Final Answer:

   [5 7 9] -> Option B

4. Quick Check:

   Sum axis=0 adds columns down rows [OK]

Hint: Sum axis=0 adds columns down rows [OK]

Common Mistakes:

• Adding rows instead of columns
• Confusing axis=0 with axis=1
• Returning flattened array instead of summed

4. What is wrong with this code snippet?

arr = np.array([[1, 2], [3, 4]])
result = np.mean(arr, axis=2)

medium

A. No error, code runs fine

B. np.mean cannot be used on arrays

C. Missing parentheses in np.array

D. axis=2 is invalid for 2D array

Solution

1. Step 1: Check array dimensions

   The array arr is 2D with shape (2,2), so valid axes are 0 and 1.

2. Step 2: Validate axis parameter

   Using axis=2 is invalid and causes an error because axis 2 does not exist.

3. Final Answer:

   axis=2 is invalid for 2D array -> Option D

4. Quick Check:

   Axis must be less than array dimensions [OK]

Hint: Axis must be less than array dimensions [OK]

Common Mistakes:

• Using axis number too large for array
• Thinking np.mean can't be used on arrays
• Ignoring error messages

5. You have a 3D NumPy array data with shape (4, 3, 2). You want to compute the sum of each 2D slice along the first axis (axis=0), resulting in a 2D array. Which code correctly does this?

hard

A. np.sum(data)

B. np.sum(data, axis=1)

C. np.sum(data, axis=0)

D. np.sum(data, axis=2)

Solution

1. Step 1: Understand array shape and axis

   Array shape is (4,3,2). Summing along axis=0 collapses the first dimension (4), resulting in shape (3,2).

2. Step 2: Apply sum along axis=0

   Using np.sum(data, axis=0) sums all 4 slices element-wise, producing a 2D array.

3. Final Answer:

   np.sum(data, axis=0) -> Option C

4. Quick Check:

   Sum axis=0 collapses first dimension to 2D [OK]

Hint: Sum axis=0 collapses first dimension in 3D arrays [OK]

Common Mistakes:

• Using wrong axis and getting wrong shape
• Summing all elements without axis
• Confusing axis numbering in 3D arrays