Defining structured dtypes in NumPy - Step-by-Step Execution

Concept Flow - Defining structured dtypes

Start: Define dtype fields

↓

Create structured dtype

↓

Create array with dtype

↓

Access fields by name

↓

Use fields for analysis

↓

End

We first define the fields and their types, then create a structured dtype. Next, we make an array using this dtype, access fields by name, and use them for analysis.

Execution Sample

NumPy

import numpy as np

# Define structured dtype
person_dtype = np.dtype([('name', 'U10'), ('age', 'i4'), ('weight', 'f4')])

# Create array
people = np.array([('Alice', 25, 55.0), ('Bob', 30, 85.5)], dtype=person_dtype)

# Access 'age' field
ages = people['age']

# Use 'ages' for analysis
mean_age = np.mean(ages)

This code defines a structured dtype for people with name, age, and weight, creates an array of people, extracts the ages, and computes the mean age.

Execution Table

Step	Action	Variable/Field	Value/Result
1	Define structured dtype	person_dtype	[('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]
2	Create array with dtype	people	[('Alice', 25, 55.0), ('Bob', 30, 85.5)]
3	Access 'age' field	ages	[25 30]
4	Use 'ages' for analysis	mean_age	27.5
5	End	-	-

💡 All steps completed: structured dtype defined, array created, fields accessed, analysis done.

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	Final
person_dtype	None	[('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]	[('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]	[('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]	[('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]	[('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]
people	None	None	[('Alice', 25, 55.0), ('Bob', 30, 85.5)]	[('Alice', 25, 55.0), ('Bob', 30, 85.5)]	[('Alice', 25, 55.0), ('Bob', 30, 85.5)]	[('Alice', 25, 55.0), ('Bob', 30, 85.5)]
ages	None	None	None	[25 30]	[25 30]	[25 30]
mean_age	None	None	None	None	27.5	27.5

Key Moments - 3 Insights

Why do we specify field names and types inside a list of tuples when defining a structured dtype?

Because each tuple defines a field name and its data type, which tells numpy how to store and access each part of the structured array. See execution_table step 1 where person_dtype is defined.

How does accessing a field like 'age' return a separate array?

Accessing a field by name extracts that column from the structured array as a normal numpy array. See execution_table step 3 where ages = people['age'] returns [25 30].

Can we perform normal numpy operations on fields extracted from structured arrays?

Yes, fields behave like normal arrays. For example, we calculate mean_age from ages in step 4.

Visual Quiz - 3 Questions

Look at the execution table, what is the value of 'people' after step 2?

A [25, 30]

B [('Alice', 25, 55.0), ('Bob', 30, 85.5)]

C [('name', '<U10'), ('age', '<i4'), ('weight', '<f4')]

D None

💡 Hint

Check the 'people' variable value in execution_table row for step 2.

At which step do we extract the 'age' field from the structured array?

A Step 1

B Step 2

C Step 3

D Step 4

💡 Hint

Look for the step where 'ages' variable is assigned in execution_table.

If we add a new field 'height' to person_dtype, which step would change?

A Steps 1 and 2

B Step 2 only

C Step 1 only

D All steps

💡 Hint

Adding a field changes dtype definition and array creation, see steps 1 and 2.

Concept Snapshot

Defining structured dtypes:
- Use np.dtype with list of (name, type) tuples
- Create arrays with this dtype
- Access fields by name like array['field']
- Fields behave like normal arrays for analysis
- Useful for mixed data types in one array

Full Transcript

We start by defining a structured dtype using numpy's dtype function with a list of tuples. Each tuple has a field name and its data type. Then, we create a numpy array using this dtype, which stores records with multiple fields. We can access any field by its name, which returns a normal numpy array of that field's values. This allows us to perform normal numpy operations on individual fields. For example, we defined a dtype for people with name, age, and weight, created an array of two people, extracted the ages, and calculated their mean age.