Intro to NumPy and pandas

numbers = list(range(1_000_000))
doubled = [n * 2 for n in numbers]      # a full Python-level loop, one iteration at a time

import numpy as np

numbers = np.arange(1_000_000)
doubled = numbers * 2                    # no explicit loop — operates on the WHOLE array at once

import numpy as np

a = np.array([1, 2, 3, 4, 5])
print(a)             # [1 2 3 4 5]
print(a.shape)        # (5,) — a 1-dimensional array of 5 elements
print(a.dtype)         # int64 — every element shares ONE data type, unlike a Python list

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

prices = np.array([19.99, 29.99, 9.99, 49.99])

with_tax = prices * 1.08            # every element multiplied at once
print(with_tax)                      # [21.5892 32.3892 10.7892 53.9892]

print(prices[prices > 20])           # [29.99 49.99] — boolean indexing: filter by a condition, no loop

import pandas as pd

df = pd.DataFrame({
    "name": ["Keyboard", "Mouse", "Monitor"],
    "price": [79.99, 24.99, 249.99],
    "in_stock": [True, True, False],
})

print(df)
#        name   price  in_stock
# 0  Keyboard   79.99      True
# 1     Mouse   24.99      True
# 2   Monitor  249.99     False

print(df["price"])              # a single column, as a Series
print(df[df["in_stock"]])       # rows where in_stock is True — boolean filtering, same idea as NumPy
print(df["price"].mean())       # 118.32333... — built-in aggregate methods
print(df["price"].sum())        # 354.97

df["price_with_tax"] = df["price"] * 1.08   # adding a new computed column, vectorised, no loop

import csv
import pandas as pd

# The csv module — you get a list of dicts, and write your own loops for everything
with open("sales.csv") as f:
    rows = list(csv.DictReader(f))
total = sum(float(row["amount"]) for row in rows)

# pandas — one line to load, built-in vectorised aggregation
df = pd.read_csv("sales.csv")
total = df["amount"].sum()

df = pd.DataFrame({
    "category": ["electronics", "electronics", "office", "office"],
    "amount": [79.99, 249.99, 4.99, 12.99],
})

print(df.groupby("category")["amount"].sum())
# category
# electronics    329.98
# office          17.98

electronics = df[df["category"] == "electronics"]
electronics["amount"] = electronics["amount"] * 1.1
# SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame

# If you intend to modify the ORIGINAL df:
df.loc[df["category"] == "electronics", "amount"] *= 1.1

# If you intend to work on an independent COPY:
electronics = df[df["category"] == "electronics"].copy()
electronics["amount"] = electronics["amount"] * 1.1   # no warning — pandas knows this is intentional

A weekly campaign-performance report reads a 2-million-row CSV using the standard csv module, computing per-campaign totals and averages with hand-written Python loops and dictionaries. The script takes 40 minutes to run, and every new metric requested by the marketing team means writing another manual loop.

totals = {}
counts = {}
with open("campaign_data.csv") as f:
    for row in csv.DictReader(f):
        campaign = row["campaign_id"]
        totals[campaign] = totals.get(campaign, 0) + float(row["spend"])
        counts[campaign] = counts.get(campaign, 0) + 1
averages = {c: totals[c] / counts[c] for c in totals}

df = pd.read_csv("campaign_data.csv")
summary = df.groupby("campaign_id")["spend"].agg(["sum", "mean", "count"])

Why the difference was dramatic, not just stylistic

The pandas version runs in roughly 3 seconds instead of 40 minutes — the vectorised groupby/agg operations are implemented in optimised C code operating on contiguous memory, exactly as described in Part 01, instead of 2 million individual Python-level dictionary lookups and updates. Just as significantly for the team's day-to-day work, adding a new requested metric became a one-line change to the .agg([...]) call instead of writing and testing an entirely new manual accumulation loop.