loc and iloc

The problem with bracket indexing

Basic bracket indexing df[mask] works for row filtering and df["col"] for column selection. But when you need to select specific rows and specific columns in one step, or edit individual cells, you need loc and iloc.

import pandas as pd

# titanic.csv ships with the course — load it from the browser file system.
df = pd.read_csv("titanic.csv")

loc: label-based selection

loc selects by label, the row index labels and column names:

# Select row at index label 0, columns "Name" and "Age"
print(df.loc[0, ["Name", "Age"]])

Output:

Name    Braund, Mr. Owen Harris
Age                        22.0
Name: 0, dtype: object

Slice by label, the endpoint is inclusive (unlike Python slicing):

# Rows 0 through 4, columns Name through Age
print(df.loc[0:4, "Name":"Age"])

Select all rows for specific columns:

print(df.loc[:, ["Name", "Survived"]].head())

Select all columns for specific rows:

print(df.loc[[0, 5, 10]])

iloc: position-based selection

iloc selects by integer position, the row/column number starting from 0:

# First row, first three columns
print(df.iloc[0, :3])

Output:

PassengerId                            1
Survived                               0
Pclass                                 3
Name: 0, dtype: object

Slice by position, the endpoint is exclusive (standard Python behavior):

# Rows 0-4 (5 rows), columns 0-2 (3 columns)
print(df.iloc[0:5, 0:3])

Select specific rows and columns:

# Rows 0, 1, 5; columns 3 (Name) and 4 (Age)
print(df.iloc[[0, 1, 5], [3, 4]])

loc vs iloc: key differences

# These are different:
df.loc[0:5]       # rows with labels 0 through 5 (inclusive) — 6 rows
df.iloc[0:5]      # rows at positions 0 through 4 (exclusive) — 5 rows

Using loc for assignment

loc is not just for reading, you can use it to edit specific cells:

# Set Age to 0 for the first passenger
df.loc[0, "Age"] = 0

# Set Fare to -1 for rows where Fare is negative
df.loc[df["Fare"] < 0, "Fare"] = 0

# Create a new column based on conditions
df.loc[df["Age"] < 18, "Category"] = "Minor"
df.loc[df["Age"] >= 18, "Category"] = "Adult"

This targeted editing is essential for data cleaning.

Practical patterns

Get a specific cell value:

# The name of the passenger at position 100
name = df.loc[100, "Name"]
print(name)

Select a range of columns:

# All rows, columns from "Name" to "Fare"
print(df.loc[:, "Name":"Fare"].head())

Conditional selection with both axes:

# Female passengers, only Name and Age columns
women = df.loc[df["Sex"] == "female", ["Name", "Age"]]
print(women.head())

Try It

From the Titanic dataset:

1. Use iloc to print the first 3 rows and first 4 columns
2. Use loc to print the Name and Fare of the passenger at index 50
3. Use loc to set the Age of the passenger at index 0 to 25

import pandas as pd

# titanic.csv ships with the course — load it from the browser file system.
df = pd.read_csv("titanic.csv")

# First 3 rows, first 4 columns
print(df.iloc[0:3, 0:4])

# Name and Fare at index 50
print(df.loc[50, ["Name", "Fare"]])

# Set Age to 25
df.loc[0, "Age"] = 25
print(df.loc[0, "Age"])

Key Takeaways

• loc selects by label (names); iloc selects by integer position
• loc slices are inclusive on both ends; iloc slices follow Python convention (exclusive end)
• loc supports assignment for targeted cell editing
• Combining row and column selection in one loc call is cleaner than chained indexing

Practice Challenge

From the Titanic dataset, use iloc to extract rows 100-109 and columns 2-5 (Pclass through Age). Then use loc to find the names of all passengers with index labels 0, 50, 100, and 500. Finally, use loc to change the Fare of passenger at index 7 to 999 and verify the change.