Stop 06 · Python for Data AnalysisTopics · 4/12

DataFrames & Series

In short

Half of early Pandas confusion is not knowing whether you're holding a table or a single column. Get this one distinction right and the error messages start making sense.

A DataFrame is Pandas' 2D table structure (rows and columns); a Series is a single column — understanding this distinction clears up most early Pandas confusion.

The jaadu holds a cricket bat and points at one column of a big scoreboard.

Think of it like this

The cricket scorecard

The full scorecard, with batters as rows and runs, balls and fours as columns, is a DataFrame. Read just the 'runs' column top to bottom and you have a Series. The batting order number on the left is the index, the label for each row.

Watch it happen

What's inside a DataFrame

DataFrame

index

columns

runs (Series)

balls (Series)

values

same index

1/4A DataFrame is the whole table: rows, columns and labels.

The big ideas

What you need to know

DataFrame = 2D

Rows and columns. Has .shape like (11, 4) and .columns listing the names.

Series = 1D

One column with an index. df["runs"] gives a Series, df[["runs"]] gives a one-column DataFrame.

.loc uses labels

df.loc["Kohli", "runs"] picks by row and column name. Slices with .loc include the end label.

.iloc uses positions

df.iloc[0, 1] is first row, second column. Works like Python lists, end excluded.

See it in code

Walk through the code

python

1import pandas as pd2 3score = pd.DataFrame(4    {"runs": [82, 45, 12], "balls": [53, 38, 9]},5    index=["Kohli", "Gill", "Pant"],6)7 8runs = score["runs"]          # Series9print(type(runs))10print(score.loc["Gill", "runs"])   # 4511print(score.iloc[0, 1])            # 5312score["strike_rate"] = score["runs"] / score["balls"] * 100

1/4Keys become column names. The index gives each row a readable label.

Build a DataFrame from a dictionary, pull out a Series, and pick cells by label and by position.

Save yourself the pain

Common mistakes

Oops

Expecting df["runs"] to behave like a table and calling DataFrame-only methods on it.

Do this instead

Check type(). Use df[["runs"]] with double brackets if you really need a DataFrame.

Oops

Mixing up .loc and .iloc after filtering, when the index is no longer 0, 1, 2.

Do this instead

Use .iloc for positions and .loc for labels, or reset_index(drop=True) after filtering.

Your mission

Your team's scorecard

~15 min

Quick check

What does df[["runs"]] return, compared to df["runs"]?

Go deeper

1Practice extracting a single column as a Series and note what changes.
2Learn to select data by label (.loc) versus by position (.iloc).
3Build a DataFrame from scratch using a Python dictionary.