Creating structured arrays in NumPy - Performance & Efficiency

Time Complexity: Creating structured arrays

O(n)

Understanding Time Complexity

We want to understand how the time needed to create structured arrays changes as the data size grows.

How does the work increase when we add more records to the array?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

import numpy as np

dtype = [('name', 'U10'), ('age', 'i4'), ('weight', 'f4')]
data = np.zeros(1000, dtype=dtype)

for i in range(1000):
    data[i] = ('Alice', i % 100, 55.0 + i * 0.1)

This code creates a structured array with 1000 records and fills each record with data.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: The for-loop that assigns values to each element in the structured array.
• How many times: Exactly once for each of the 1000 records.

How Execution Grows With Input

As the number of records increases, the time to fill the array grows proportionally.

Input Size (n)	Approx. Operations
10	About 10 assignments
100	About 100 assignments
1000	About 1000 assignments

Pattern observation: Doubling the number of records roughly doubles the work done.

Final Time Complexity

Time Complexity: O(n)

This means the time to create and fill the structured array grows linearly with the number of records.

Common Mistake

[X] Wrong: "Creating a structured array is instant and does not depend on size."

[OK] Correct: Each record must be assigned data, so the time grows with the number of records.

Interview Connect

Understanding how data creation scales helps you explain performance in real data tasks, a useful skill in interviews and projects.

Self-Check

"What if we used vectorized assignment instead of a for-loop? How would the time complexity change?"

Practice

1. What is the main purpose of creating a structured array in numpy?

easy

A. To create arrays with only one data type

B. To store data with different types in named fields within one array

C. To speed up numerical calculations on large arrays

D. To visualize data using plots

Solution

1. Step 1: Understand structured arrays

   Structured arrays allow storing mixed data types in one array with named fields.

2. Step 2: Compare options

   Only To store data with different types in named fields within one array correctly describes this purpose; others describe unrelated features.

3. Final Answer:

   To store data with different types in named fields within one array -> Option B

4. Quick Check:

   Structured arrays = mixed types + named fields [OK]

Hint: Structured arrays hold mixed data types with names [OK]

Common Mistakes:

• Thinking structured arrays only hold one data type
• Confusing structured arrays with plotting functions
• Assuming structured arrays speed up all calculations

2. Which of the following is the correct way to define a structured array dtype with fields 'name' (string) and 'age' (integer)?

easy

A. dtype = [('name', 'U10'), ('age', 'i4')]

B. dtype = ['name': str, 'age': int]

C. dtype = {'name': 'string', 'age': 'int'}

D. dtype = [('name', str), ('age', float)]

Solution

1. Step 1: Recall dtype syntax for structured arrays

   Structured array dtypes are defined as a list of tuples: (field_name, data_type).

2. Step 2: Check each option

   dtype = [('name', 'U10'), ('age', 'i4')] uses correct tuple syntax with numpy string and integer types. Others use invalid syntax or wrong types.

3. Final Answer:

   dtype = [('name', 'U10'), ('age', 'i4')] -> Option A

4. Quick Check:

   Structured dtype = list of (name, type) tuples [OK]

Hint: Use list of (field, type) tuples for dtype [OK]

Common Mistakes:

• Using dictionary syntax instead of list of tuples
• Using Python types instead of numpy dtype strings
• Mixing float type for integer fields

3. What will be the output of the following code?

import numpy as np
dtype = [('id', 'i4'), ('score', 'f4')]
data = np.array([(1, 9.5), (2, 8.0)], dtype=dtype)
print(data['score'])

medium

A. Error: KeyError

B. [1 2]

C. [(1, 9.5) (2, 8.0)]

D. [9.5 8. ]

Solution

1. Step 1: Understand structured array creation

   The array has fields 'id' (int) and 'score' (float), with two records.

2. Step 2: Access the 'score' field

   Accessing data['score'] returns an array of the 'score' values: [9.5, 8.0].

3. Final Answer:

   [9.5 8. ] -> Option D

4. Quick Check:

   Accessing field returns array of that column [OK]

Hint: Access fields by name to get column arrays [OK]

Common Mistakes:

• Expecting full records instead of single field
• Confusing field names causing KeyError
• Thinking output is list of tuples

4. Identify the error in the following code that tries to create a structured array:

import numpy as np
dtype = [('name', 'U5'), ('age', 'i4')]
data = np.array([('Alice', 25), ('Bob', 30)], dtype=dtype)
print(data['age'])

medium

A. The dtype definition is missing field names

B. The tuple elements should be lists, not tuples

C. No error; code runs correctly

D. The string length 'U5' is too short for 'Alice'

Solution

1. Step 1: Check dtype and data compatibility

   'U5' means Unicode string of length 5, 'Alice' has 5 characters, so it fits.

2. Step 2: Verify data structure and code correctness

   Data is a list of tuples matching dtype fields; code runs without error and prints ages.

3. Final Answer:

   No error; code runs correctly -> Option C

4. Quick Check:

   String length matches field size; tuples allowed [OK]

Hint: Check string length matches longest string [OK]

Common Mistakes:

• Assuming string length too short causes error
• Thinking tuples are invalid for data input
• Believing dtype must be dictionary

5. You have a list of employee data: [('John', 28, 50000), ('Jane', 32, 60000), ('Doe', 24, 45000)]. How do you create a structured array with fields 'name' (string, max 10 chars), 'age' (int), and 'salary' (float) to store this data?

hard

A. dtype = [('name', 'U10'), ('age', 'i4'), ('salary', 'f8')]; np.array(data, dtype=dtype)

B. dtype = [('name', str), ('age', int), ('salary', float)]; np.array(data)

C. dtype = {'name': 'U10', 'age': 'i4', 'salary': 'f8'}; np.array(data)

D. dtype = [('name', 'S10'), ('age', 'i4'), ('salary', 'f4')]; np.array(data)

Solution

1. Step 1: Define correct dtype for fields

   Use list of tuples with field names and numpy types: 'U10' for Unicode string max 10 chars, 'i4' for int, 'f8' for float64.

2. Step 2: Create structured array with data and dtype

   Pass data and dtype to np.array to create structured array storing all fields correctly.

3. Final Answer:

   dtype = [('name', 'U10'), ('age', 'i4'), ('salary', 'f8')]; np.array(data, dtype=dtype) -> Option A

4. Quick Check:

   Use list of (field, type) tuples and pass dtype [OK]

Hint: Use Unicode string 'U10' for names, correct numeric types [OK]

Common Mistakes:

• Using Python types instead of numpy dtype strings
• Using dictionary instead of list of tuples for dtype
• Choosing byte string 'S10' instead of Unicode 'U10'