Functions and NumPy Basics

Functions

A function is a named, reusable block of code. You have already used built-in ones (print, len, sum). Now you will write your own.

def calculate_revenue(units, price, discount=0):
    '''Return revenue after an optional discount (0.1 = 10%).'''
    revenue = units * price * (1 - discount)
    return round(revenue, 2)

calculate_revenue(10, 4.5)        # 45.0
calculate_revenue(10, 4.5, 0.1)   # 40.5

• def starts the definition; the indented lines are the body.
• Parameters (units, price, discount) are the inputs. discount=0 is a default used if you leave it out.
• return sends a result back. A function without return gives back None.
• The text in triple quotes is a docstring explaining what the function does.

Why bother? If you clean data the same way every month, a function lets you do it with one line, and if you find a mistake you fix it in one place. A good rule: if you copy and paste code a third time, make it a function.

NumPy Basics

NumPy (Numerical Python) adds the array: a grid of numbers that Python can calculate on very quickly. Import it with the standard nickname:

import numpy as np
prices = np.array([19.99, 5.49, 12.00, 7.25])

Vectorized math means one operation applies to every element without a loop:

prices * 1.1            # add 10% to every price
prices + 1              # add 1 to every price
prices.mean(), prices.max(), prices.sum()
np.median(prices), prices.std()

Boolean filtering picks out elements that meet a condition:

prices > 10              # array([ True, False,  True, False])
prices[prices > 10]      # array([19.99, 12.  ])

Arrays can have two dimensions (rows and columns). arr.shape tells you their size, such as (4, 3) for 4 rows and 3 columns.

Why it matters: pandas, the library you learn next week, is built on top of NumPy. The same ideas (vectorized math and boolean filters) are how you will work with whole columns of data.