Learn NumPy

What Is NumPy?

NumPy, short for Numerical Python, is an open-source library that adds support for large, multi-dimensional arrays and matrices, along with a huge collection of mathematical functions to operate on them efficiently. It is the foundation that most of the scientific Python ecosystem - including pandas, SciPy, scikit-learn, and TensorFlow - is built on top of.

The ndarray: NumPy's Core Object

At the heart of NumPy is the ndarray (n-dimensional array). Unlike a Python list, which stores references to scattered objects, an ndarray stores its data in one contiguous block of memory, and every element must share the same data type.

• Homogeneous - every element has the same data type (e.g. all int64 or all float64)
• Fixed size - the total number of elements is set when the array is created
• Multi-dimensional - a single ndarray can represent a vector, a matrix, or an n-dimensional tensor
• Contiguous memory - elements are packed together rather than scattered like list references
• Vectorized - mathematical operations apply to the whole array at once, without an explicit Python loop

Why Is NumPy Faster Than Python Lists?

A Python list stores pointers to individual Python objects, each carrying its own type information and reference count. Every time you loop over a list, Python has to check each element's type and unbox it before doing any math. An ndarray skips all of that: because every element is the same known type packed into a single memory block, NumPy can hand the whole array to precompiled, low-level C loops (and even CPU vector instructions) that process it in one shot instead of one Python bytecode step at a time.

import numpy as np

py_list = [1, 2, 3, 4]
np_array = np.array([1, 2, 3, 4])

print(type(py_list))    # <class 'list'>
print(type(np_array))   # <class 'numpy.ndarray'>
print(np_array.dtype)   # int64

import numpy as np

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

# Every element is squared in one step - no loop required
squared = numbers ** 2
print(squared)   # [ 1  4  9 16 25]

import numpy as np
import time

numbers = list(range(1_000_000))
np_numbers = np.array(numbers)

start = time.time()
total = sum(numbers)
print('Python list sum time:', time.time() - start)

start = time.time()
total = np_numbers.sum()
print('NumPy array sum time:', time.time() - start)