NumPy Array Creation

Creating arrays from Python objects

From a list

from-list

import numpy as np
 
arr = np.array([1, 2, 3, 4])
print(arr)
print(type(arr))

From a nested list (2D)

from-nested

matrix = np.array([
    [1, 2, 3],
    [4, 5, 6]
])
print(matrix)
print(matrix.shape)   # (2, 3)

From a tuple

from-tuple

arr = np.array((10, 20, 30))
print(arr)

Using built-in constructors

np.zeros()

Create an array filled with zeros.

zeros

arr = np.zeros((2, 3))
print(arr)

np.ones()

ones

arr = np.ones((3, 2))
print(arr)

np.full()

Create an array filled with a constant value.

full

arr = np.full((2, 2), 7)
print(arr)

np.eye() (identity matrix)

eye

I = np.eye(3)
print(I)

np.arange() (range with step)

Similar to Python range(), but returns a NumPy array.

arange

arr = np.arange(0, 10, 2)
print(arr)  # [0 2 4 6 8]

np.linspace() (even spacing)

Creates num values between start and stop.

linspace

arr = np.linspace(0, 1, 5)
print(arr)  # [0.   0.25 0.5  0.75 1.  ]

Specifying dtype during creation

dtype

arr = np.array([1, 2, 3], dtype=np.float64)
print(arr)
print(arr.dtype)

From the book

McKinney warns that numpy.empty does not return an array of zeros — it hands back a chunk of raw, uninitialized memory that may contain “garbage” values left over from something else. Only reach for np.empty when you plan to immediately fill every element yourself; otherwise use np.zeros.

Choosing the right constructor

diagram Which array-creation function? mermaid

  flowchart TD
  A["Need an array"] --> B{"Already have data
(list/tuple)?"}
  B -- "Yes" --> C["np.array(data)"]
  B -- "No" --> D{"What kind of
placeholder?"}
  D -- "All zeros" --> E["np.zeros(shape)"]
  D -- "All ones" --> F["np.ones(shape)"]
  D -- "One constant value" --> G["np.full(shape, value)"]
  D -- "Identity matrix" --> H["np.eye(n)"]
  D -- "Integer sequence" --> I["np.arange(start, stop, step)"]
  D -- "Evenly spaced floats" --> J["np.linspace(start, stop, num)"]

A decision flow for picking the right NumPy constructor based on what data you already have.

Quick recap: which function to use?

• np.array(...) → convert existing data (lists)
• np.zeros(shape) / np.ones(shape) → initialize arrays
• np.full(shape, value) → constant arrays
• np.eye(n) → identity matrix
• np.arange(start, stop, step) → integer sequences
• np.linspace(start, stop, num) → precise evenly spaced floats

Next

Continue to: NumPy Data Types (dtypes) to learn how dtypes affect memory, performance, and numeric precision.

🧪 Try It Yourself

Exercise 1 – Zeros, Ones, and Full

pch.exerciseTag pch.exerciseTitle

Exercise 2 – arange vs linspace

pch.exerciseTag pch.exerciseTitle

Exercise 3 – Identity Matrix

pch.exerciseTag pch.exerciseTitle