Functions — Definition, Arguments & Scope

def greet(name):
    return f"Hello, {name}!"

result = greet("Priya")  # "Hello, Priya!"

Functions encapsulate logic so it can be reused. In data work, you write functions like clean_phone(number), standardise_city(name), or compute_kpi(df) that are applied across columns or DataFrames.

def describe(name, role="Analyst", city="Noida"):
    return f"{name} is a {role} in {city}"

describe("Rahul")                         # uses defaults
describe("Priya", city="Delhi")           # keyword arg
describe("Aman", "Engineer", "Mumbai")    # positional

Default arguments make functions flexible. In Pandas functions like pd.read_csv(), almost everything is a keyword argument with a sensible default. Understanding this helps you read library docs.

def total(*args):            # any number of positional args
    return sum(args)
total(1, 2, 3, 4)   # 10

def info(**kwargs):           # any number of keyword args
    for k, v in kwargs.items():
        print(k, v)
info(name="Rahul", age=28)

*args collects extra positional arguments into a tuple; **kwargs collects extra keyword arguments into a dict. Used in wrapper functions and when the number of parameters is unknown — common in analytics utility functions.

# Regular function:
def square(x):
    return x ** 2

# Lambda equivalent:
square = lambda x: x ** 2
square(5)  # 25

# Common in Pandas:
df["salary_lakhs"] = df["salary"].apply(lambda x: round(x / 100000, 2))

Lambda functions are anonymous one-liners. They are most useful inside .apply(), map(), sorted(), and filter(). For anything more than one line, use a regular def — readability matters.

x = 10  # global

def modify():
    x = 20  # local — does not affect global x
    print(x)  # 20

modify()
print(x)  # 10 — unchanged

# To modify global:
def modify_global():
    global x
    x = 20

Variables inside a function are local by default. Avoid global variables in data scripts — they make code hard to debug. Pass values as arguments and return results instead.