Working with Columns

Being Specific About What You Select

In the last lesson, you learned to select columns with double brackets. That’s the foundation. But real analysis requires more than just pulling raw columns out of a DataFrame — you’ll need to rename them, compute new ones, and clean up duplicates.

Let’s build those skills one at a time.

Column Order Is Up to You

techcorp_workforce[["department", "first_name", "salary"]]

This seems trivial, but it matters when you’re building reports or exporting data. Put the most important columns first.

Renaming Columns

DataFrame column names such as employee_id or joining_date make sense to developers. But when you’re building a report for stakeholders, you want something more readable.

The rename() method lets you change column names in the output:

cols = techcorp_workforce[["first_name", "joining_date"]]
cols.rename(columns={
    "first_name": "First Name",
    "joining_date": "Hire Date"
})

You pass a dictionary to the columns parameter: keys are old names, values are new names. Only the columns you list get renamed — everything else stays the same.

Column Aliases

Creating New Columns with Arithmetic

One of the most useful things you can do with DataFrames is create new columns from existing ones. You just assign to a new column name using standard math operators:

techcorp_workforce["monthly_salary"] = techcorp_workforce["salary"] / 12

This creates a new column called monthly_salary by dividing the salary column by 12. The operation applies to every row automatically — no loops needed.

All the standard operators work:

• + addition — salary + 5000 adds a bonus
• - subtraction — salary - tax calculates net pay
• * multiplication — price * quantity gets a total
• / division — salary / 12 converts annual to monthly

Computed Columns

Arithmetic on Columns

Combining Text Columns

Sometimes you need to merge text from multiple columns — like creating a full name from first_name and last_name. You concatenate strings with the + operator:

techcorp_workforce["full_name"] = (
    techcorp_workforce["first_name"] + " " + techcorp_workforce["last_name"]
)

The ' ' in the middle adds a space between the two names. Without it, you’d get JohnSmith instead of John Smith.

Building Full Names

Removing Duplicates

Sometimes you want to know what values exist in a column, not see every row. The drop_duplicates() method removes duplicate rows from your results.

Let’s look at the orders table:

Without removing duplicates, you’d see repeated order details:

Unique Order Products

Deduplicating on Multiple Columns

When you select multiple columns before calling drop_duplicates(), it removes rows where the entire combination is duplicated:

This returns unique customer-product pairs. If customer 3 ordered Shoes twice, you’d only see that combination once.

A Quick Note on Data Types

Every column in a DataFrame has a data type. You saw dtypes in the last lesson. Here’s what the main types mean:

• object — text (strings). Names, emails, categories.
• int64 — whole numbers. IDs, counts, ages.
• float64 — decimal numbers. Prices, percentages, measurements.
• datetime64 — dates and timestamps. When things happened.
• bool — True/False values. Flags, yes/no fields.

Why does this matter? Because you can’t do math on text, and you can’t search for string patterns in numbers. If something isn’t working, check dtypes first — a column you think is numeric might actually be stored as text.

Common Mistakes

Misspelling column names

# KeyError: "firstname"
techcorp_workforce["firstname"]

# The column is actually called "first_name"
techcorp_workforce["first_name"]

You’ll get a KeyError if the name doesn’t match exactly. Check df.columns if you’re not sure.

Forgetting the double brackets

# This returns a Series, not a DataFrame
techcorp_workforce["first_name"]

# This returns a DataFrame
techcorp_workforce[["first_name"]]

If a method complains about getting a Series when it expects a DataFrame, this is probably why.

Using the wrong quotes

# Without quotes, Python thinks these are variables
techcorp_workforce[[first_name, last_name]]  # NameError!

# Column names are always strings — use quotes
techcorp_workforce[["first_name", "last_name"]]  # correct

Renaming with a New Dataset

Let’s practice with the restaurant inspections data:

Unique Values in a Column

Key Takeaways

• Column order in your selection determines the order in your output. Put the most important columns first.
• Use rename(columns={...}) to give columns readable names.
• Create new columns with arithmetic: df['new'] = df['col'] * 2. The operation applies to every row automatically.
• Concatenate text with +. Remember to convert non-string columns with .astype(str) first.
• drop_duplicates() removes duplicate rows. It applies to the full row, not just the first column.
• When something isn’t working, check df.dtypes — the column might not be the type you expect.

What’s Next

Now you know how to select, rename, compute, and deduplicate columns. But so far, you’re getting all rows. In the next lesson, you’ll learn to filter data with Boolean indexing — which is when pandas starts to feel actually powerful.