NumPy Arrays — Fundamentals

import numpy as np
arr = np.array([1, 2, 3, 4, 5])
arr * 2   # [2, 4, 6, 8, 10] — vectorised, no loop needed

NumPy arrays store data in contiguous memory blocks and use C-level operations — no Python interpreter overhead per element. Operations on 1 million elements are 10-100x faster than a Python loop.

np.array([1, 2, 3])         # from list
np.zeros((3, 4))             # 3x4 array of 0s
np.ones((2, 3))              # 2x3 array of 1s
np.arange(0, 10, 2)         # [0, 2, 4, 6, 8]
np.linspace(0, 1, 5)        # [0.0, 0.25, 0.5, 0.75, 1.0]
np.random.randint(0, 100, (3, 3))  # random 3x3

arange() is like Python range() but returns an array. linspace() creates evenly spaced values — useful for generating axis labels, bins, or simulation inputs.

a = np.array([10, 20, 30])
b = np.array([1, 2, 3])

a + b   # [11, 22, 33]
a - b   # [9, 18, 27]
a * b   # [10, 40, 90]
a / b   # [10.0, 10.0, 10.0]
a ** 2  # [100, 400, 900]

All arithmetic operations are element-wise and vectorised. No loops needed. This is how Pandas column arithmetic works internally.

arr = np.array([[1, 2, 3],
                [4, 5, 6],
                [7, 8, 9]])

arr[0]        # [1, 2, 3]  — first row
arr[:, 1]     # [2, 5, 8]  — second column
arr[1, 2]     # 6  — row 1, col 2
arr[0:2, 1:]  # [[2,3],[5,6]]

The syntax is arr[rows, cols]. This is exactly how Pandas .iloc[] works — it uses NumPy indexing underneath. Master this and .iloc[] becomes intuitive.

arr = np.array([10, 20, 30, 40, 50])
np.mean(arr)    # 30.0
np.median(arr)  # 30.0
np.std(arr)     # standard deviation
np.var(arr)     # variance
np.sum(arr)     # 150
np.min(arr)     # 10
np.max(arr)     # 50
np.percentile(arr, 75)  # 40.0

These are the building blocks of EDA. Pandas descriptive stats (.describe()) calls these NumPy functions internally. Knowing both layers helps you understand what .describe() actually computes.