NumPy - Creating Structured Arrays

---

---

Introduction to Structured Arrays

Structured arrays in NumPy allow for the representation of data with different types and sizes in a single array. Each element in a structured array can be a record with multiple fields, each field having its own data type.

This is similar to having a table where each row represents a record with various attributes. The key points are as follows −

• Each element (record) in a structured array can have multiple fields.
• Fields can have different data types (e.g., integers, floats, strings).
• Structured arrays are useful for representing complex data structures.

Defining a Structured Array

The first step for creating a structured array is to define the data type for the structured array. This is done using a NumPy dtype object, which specifies the names and types of the fields in the array as shown in the following example −

import numpy as np

# Define the data type for the structured array
dtype = np.dtype([('name', 'U10'), ('age', 'i4'), ('height', 'f4')])

Here, dtype is a structured data type with three fields −

• name: A string of up to 10 characters ('U10')
• age: An integer ('i4')
• height: A floating-point number ('f4')

Creating the Structured Array

Once you have defined the "dtype", you can create the structured array by passing the dtype to the np.array() function. Following is the syntax −

numpy.array(
   object, 
   dtype=None, 
   copy=True, 
   order='K', 
   subok=False, 
   ndmin=0
)

Where,

• object: This is the input data that you want to convert into a NumPy array. It can be a list, tuple, or any other sequence-like structure.
• dtype (Optional): This specifies the desired data type for the array. If not provided, NumPy will infer the data type from the input data. For example, you can use 'int32', 'float64', 'str', etc.
• copy (Optional): If True (default), a new array is created. If False, a new array is created only if necessary (i.e., if the input object is not already an array). If False, np.array may return a view of the original array if possible.
• order (Optional): This specifies the memory layout order. It can be 'C' for row-major (C-style) order, 'F' for column-major (Fortran-style) order, or 'K' for the order as found in the input. Default is 'K'.
• subok (Optional): If True, a subclass of ndarray will be used if the input is a subclass. Default is False, meaning the returned array will always be an instance of ndarray.
• ndmin (Optional): This specifies the minimum number of dimensions that the resulting array should have. For example, setting ndmin=2 ensures that the result is at least a 2-dimensional array.

Example

In the following example, we are defining a structured array with a specified dtype that includes fields for "name", "age", and "height". We then create the array with data matching this structure and print the resulting structured array −

import numpy as np

# Define the dtype with field names and data types
dtype = [('name', 'U10'), ('age', 'i4'), ('height', 'f4')]

# Create data consistent with the dtype
data = [('Alice', 30, 5.6), ('Bob', 25, 5.8), ('Charlie', 35, 5.9)]

# Create the structured array
structured_array = np.array(data, dtype=dtype)
print("Structured Array:\n", structured_array)

Following is the output obtained −

Structured Array:
[('Alice', 30, 5.6) ('Bob', 25, 5.8) ('Charlie', 35, 5.9)]

Create Structured Arrays with Different Data Types

You can create structured arrays with various data types, including strings, integers, and floats, depending on the needs of your application.

In the context of structured arrays −

• Strings: You can store text data in structured arrays. For example, a field for names might use a string data type.
• Integers: Numerical data, such as ages or counts, can be stored as integers. This might include data like age or quantity, which are whole numbers.
• Floats: For decimal numbers or real numbers, you can use float data types. This is useful for measurements or any value requiring precision, such as height or weight.

Example

In the following example, we are creating a structured array with a data type (dtype) that includes mixed data types: integers for IDs, strings for names, and floating-point numbers for scores −

import numpy as np

# Define a dtype with mixed data types
dtype = [('id', 'i4'), ('name', 'U15'), ('score', 'f8')]
data = [(1, 'Alice', 88.5), (2, 'Bob', 91.2), (3, 'Charlie', 85.4)]

# Create the structured array
structured_array = np.array(data, dtype=dtype)
print("Structured Array with Mixed Data Types:\n", structured_array)

This will produce the following result −

Structured Array with Mixed Data Types:[(1, 'Alice', 88.5) (2, 'Bob', 91.2) (3, 'Charlie', 85.4)]

Create Structured Arrays Using List of Tuples

You can define the dtype and create the structured array using a list of tuples, where each tuple represents a field. Each tuple contains two elements: the first element is the name of the field, and the second element is the data type of that field.

Example

In the following example, we are defining a structured array with fields for "name", "age", and "height" using a specified dtype. We then create this array with corresponding data −

import numpy as np

# Define the dtype
dtype = [('name', 'U10'), ('age', 'i4'), ('height', 'f4')]

# Define the data
data = [('Alice', 30, 5.6), ('Bob', 25, 5.8), ('Charlie', 35, 5.9)]

# Create the structured array
structured_array = np.array(data, dtype=dtype)

print("Structured Array:\n", structured_array)

Following is the output obtained −

Structured Array:
[('Alice', 30, 5.6) ('Bob', 25, 5.8) ('Charlie', 35, 5.9)]

Create Structured Arrays Using Dictionary

Alternatively, you can define the data and dtype using a dictionary to clearly specify the names and types of fields. Each key in the dictionary represents a field name, and the value associated with each key defines the data type of that field.

Example

In this example, we are defining the dtype for a structured array using a dictionary format to specify fields for "name", "age", and "height". We then create and display this structured array with the corresponding data, organizing it into a format that supports multiple data types within each record −

import numpy as np

# Define the dtype using a dictionary
dtype = np.dtype([('name', 'U10'), ('age', 'i4'), ('height', 'f4')])

# Define the data
data = [('Alice', 30, 5.6), ('Bob', 25, 5.8), ('Charlie', 35, 5.9)]

# Create the structured array
structured_array = np.array(data, dtype=dtype)

print("Structured Array from Dictionary:\n", structured_array)

This will produce the following result −

Structured Array from Dictionary:
[('Alice', 30, 5.6) ('Bob', 25, 5.8) ('Charlie', 35, 5.9)]