Pandas Series – Part 2: Common Gotchas Around Indexing

This is Part 2 of the Pandas Series, where we explore common Pandas gotchas that can level up your interview answers and your daily data work.

The Core Concepts

1. What is an Index?
Think of the DataFrame index not as a row number, but as a set of labels or addresses for your rows. Like a dictionary key, it's optimized for fast lookups. When you write df[df['column'] == 'value'], Pandas has to scan the entire column. When you use a well-structured index with df.loc['label'], Pandas can jump directly to the data.

1. set_index() and reset_index() These are your primary tools for shaping your index.

df.set_index('column_name'): Promotes one or more columns to become the index.

df.reset_index(): Demotes the index level(s) back to being regular columns.

Use these wisely — a well-structured index can cut query time dramatically, especially on large datasets.

3. The MultiIndex (Hierarchical Indexing)
This is the "power move." A MultiIndex allows you to have multiple levels of indexing. Imagine a book's table of contents with Chapters, then Sections within Chapters. This structure lets you "drill down" to your data with incredible speed and precision.

Let's look at an example. A DataFrame with an index on (store, product):

		sales
store	product
Store_A	Apples	100
	Oranges	150
Store_B	Apples	80
	Bananas	120
	Oranges	90

With this structure, you can easily select:

All data for Store_A:
df.loc['Store_A']

Data for Apples in Store_B:
df.loc[('Store_B', 'Apples')]

Pro Tip: For range-based slicing (like with dates), the index must be sorted for optimal performance. You can do this with df.sort_index().

And that’s it for Day 2 👏
Indexing is one of those Pandas topics most people use without really understanding — but once you do, you unlock serious performance and clarity.

What’s your favourite indexing trick or gotcha you’ve run into while working with Pandas?