Level 1 · Compute line revenue

Arithmetic on whole columns

When you multiply two columns, pandas multiplies them row by row and returns a new Series:

order_items["quantity"] * order_items["unit_price"]

No loop needed. This is called vectorised arithmetic, and it’s both shorter and far faster than looping in Python. You can mix columns and plain numbers freely: order_items["unit_price"] * 1.08 applies the same number to every row.

Know your units

Look at discount_pct before using it:

order_items["discount_pct"].value_counts()
# 0.0, 10.0, 15.0, 20.0, 25.0

It’s stored in percentage points: 10.0 means 10% off. So the share the customer pays is 1 - discount_pct / 100. If you wrote 1 - discount_pct, a 10% discount would make revenue negative. Other datasets store the same idea as a fraction (0.10); always check a few values.

Line revenue is therefore:

order_items["quantity"] * order_items["unit_price"] * (1 - order_items["discount_pct"] / 100)

Two ways to add the column

The direct way assigns to a new column name, which modifies the DataFrame in place:

items = order_items.copy()
items["line_revenue"] = (
    items["quantity"] * items["unit_price"] * (1 - items["discount_pct"] / 100)
).round(2)

assign returns a new DataFrame with the column added and leaves the original alone:

items = order_items.assign(
    line_revenue=(
        order_items["quantity"] * order_items["unit_price"]
        * (1 - order_items["discount_pct"] / 100)
    ).round(2)
)

assign fits nicely into method chains. Inside a chain, where the intermediate DataFrame has no name, pass a function that receives it:

order_items.assign(gross=lambda d: d["quantity"] * d["unit_price"])

Both styles are common. The important part is to avoid chained assignment, such as order_items[order_items["quantity"] > 0]["line_revenue"] = .... That writes to a temporary copy, not to order_items. To change only some rows, use one .loc call: items.loc[items["quantity"] <= 0, "line_revenue"] = 0.

Rounding

.round(2) on a Series rounds every value. For money, round the column you report, and keep in mind that the sum of rounded lines can differ by a cent or two from the rounded sum of unrounded lines. Finance teams usually define which one is official; when in doubt, ask.

Other new-column patterns

orders["has_coupon"] = orders["coupon_code"].notna()            # boolean flag
orders["order_year"] = orders["order_date"].dt.year             # part of a date
products["margin"] = products["unit_price"] - products["unit_cost"]   # NaN where cost is missing

Arithmetic with a missing value gives a missing value, so margin is NaN for the nine products without a cost. That’s correct: you don’t know their margin.

Common mistakes

• Treating discount_pct as a fraction.
• Using products.unit_price (today’s list price) instead of the line’s unit_price (price actually paid). Prices rose on some products in March 2025, so the two differ on older lines.
• Expecting order_items.assign(...) to change order_items without storing the result.