Selecting Rows with iloc

What is iloc?

iloc selects by position (numeric index).

Think of it as selecting by row number, starting from 0.

code.pyPython

import pandas as pd

df = pd.DataFrame({
    'Name': ['John', 'Sarah', 'Mike', 'Emma'],
    'Age': [25, 30, 28, 32],
    'City': ['NYC', 'LA', 'Chicago', 'Miami']
})

print(df)

Select Single Row

code.pyPython

row = df.iloc[0]
print(row)

Gets first row (position 0).

Returns Series.

Select Multiple Rows

code.pyPython

rows = df.iloc[[0, 2]]
print(rows)

Gets rows at positions 0 and 2.

Returns DataFrame.

Slicing Rows

code.pyPython

subset = df.iloc[0:2]
print(subset)

Important: Excludes endpoint! Gets rows 0 and 1 only.

Same as Python list slicing.

Negative Indexing

code.pyPython

last_row = df.iloc[-1]
print(last_row)

last_two = df.iloc[-2:]
print(last_two)

-1 is last row, -2 is second-to-last.

Select Rows and Columns

code.pyPython

value = df.iloc[0, 0]
print(value)

Gets row 0, column 0.

Multiple rows and columns:

code.pyPython

subset = df.iloc[[0, 1], [0, 2]]
print(subset)

Gets:

Rows 0, 1
Columns 0, 2

Slicing Rows and Columns

code.pyPython

subset = df.iloc[0:2, 0:2]
print(subset)

Gets:

Rows 0, 1
Columns 0, 1

All Rows, Specific Columns

code.pyPython

first_two_cols = df.iloc[:, 0:2]
print(first_two_cols)

: means all rows.

Specific Rows, All Columns

code.pyPython

first_two_rows = df.iloc[0:2, :]
print(first_two_rows)

Can omit second ::

code.pyPython

first_two_rows = df.iloc[0:2]

Step in Slicing

code.pyPython

every_other = df.iloc[::2]
print(every_other)

Gets rows 0, 2, 4...

Custom step:

code.pyPython

subset = df.iloc[0:4:2]

Gets rows 0 and 2.

Practice Example

The scenario: Process sales data by position.

code.pyPython

import pandas as pd

sales = pd.DataFrame({
    'Date': ['2024-01-01', '2024-01-02', '2024-01-03',
             '2024-01-04', '2024-01-05', '2024-01-06'],
    'Product': ['Laptop', 'Phone', 'Tablet', 'Monitor', 'Keyboard', 'Mouse'],
    'Quantity': [5, 10, 7, 3, 15, 20],
    'Price': [999, 599, 399, 299, 75, 25]
})

print("All data:")
print(sales)
print()

print("First sale:")
print(sales.iloc[0])
print()

print("Last sale:")
print(sales.iloc[-1])
print()

print("First 3 sales:")
print(sales.iloc[0:3])
print()

print("Last 2 sales:")
print(sales.iloc[-2:])
print()

print("Every other sale:")
print(sales.iloc[::2])
print()

print("First 3 rows, first 2 columns:")
print(sales.iloc[0:3, 0:2])
print()

print("Specific rows and columns:")
specific = sales.iloc[[0, 2, 4], [1, 3]]
print(specific)
print()

print("Calculate revenue for first 3:")
first_three = sales.iloc[0:3]
first_three_copy = first_three.copy()
first_three_copy['Revenue'] = first_three_copy['Quantity'] * first_three_copy['Price']
print(first_three_copy)

Random Rows

code.pyPython

import numpy as np

random_indices = np.random.choice(len(df), 2, replace=False)
random_rows = df.iloc[random_indices]
print(random_rows)

First and Last N Rows

code.pyPython

first_3 = df.iloc[:3]
last_3 = df.iloc[-3:]

print("First 3:", first_3)
print("Last 3:", last_3)

Middle Rows

code.pyPython

middle = df.iloc[1:-1]
print(middle)

Skips first and last.

Setting Values

code.pyPython

df.iloc[0, 0] = 'Johnny'
print(df)

Multiple values:

code.pyPython

df.iloc[0:2, 1] = 99
print(df)

loc vs iloc

loc (label-based)

code.pyPython

df.loc[0]  # Row with index label 0
df.loc[0:2]  # Rows 0, 1, AND 2 (includes end)
df.loc[:, 'Name']  # Column named 'Name'
df.loc[df['Age'] > 30]  # Boolean indexing

iloc (position-based)

code.pyPython

df.iloc[0]  # First row (position 0)
df.iloc[0:2]  # Rows 0, 1 (excludes 2)
df.iloc[:, 0]  # First column
# No boolean indexing with iloc

When to Use iloc

Use iloc when:

Selecting by position
Don't care about index labels
Want first/last N rows
Slicing with numeric positions

Use loc when:

Selecting by label
Boolean conditions
Custom index names
Column names

Custom Index Example

code.pyPython

df_custom = pd.DataFrame({
    'Value': [10, 20, 30]
}, index=['A', 'B', 'C'])

print(df_custom.loc['B'])  # Row with label 'B'
print(df_custom.iloc[1])  # Second row (same result)

Both get same row, different methods.

Combining iloc with Operations

code.pyPython

top_3_avg = df.iloc[0:3]['Age'].mean()
print("Average age of first 3:", top_3_avg)

last_5_sum = df.iloc[-5:]['Salary'].sum()
print("Total salary of last 5:", last_5_sum)

Key Points to Remember

iloc selects by position (0, 1, 2...), not by label.

iloc[0:2] excludes endpoint - gets rows 0 and 1 only (like Python lists).

Negative indexing works: iloc[-1] is last row, iloc[-2] is second-to-last.

iloc[row, col] format where both are numeric positions.

Cannot use boolean conditions directly with iloc. Use loc for that.

Common Mistakes

Mistake 1: Confusing loc and iloc

code.pyPython

df.iloc['Name']  # Error! iloc uses positions
df.loc[:, 'Name']  # Correct for column name
df.iloc[:, 0]  # Correct for first column

Mistake 2: Expecting inclusive slicing

code.pyPython

df.iloc[0:3]  # Gets rows 0, 1, 2 (excludes 3)
df.loc[0:3]  # Gets rows 0, 1, 2, 3 (includes 3)

Mistake 3: Boolean indexing with iloc

code.pyPython

df.iloc[df['Age'] > 30]  # Error!
df.loc[df['Age'] > 30]  # Correct

Mistake 4: Out of range

code.pyPython

df.iloc[100]  # Error if only 10 rows
# Check length first:
if len(df) > 100:
    row = df.iloc[100]

Mistake 5: Modifying slice without copy

code.pyPython

subset = df.iloc[0:3]
subset['New'] = 1  # May affect original!

subset = df.iloc[0:3].copy()  # Safe

What's Next?

You now know how to use iloc for position-based selection. Next, you'll learn about filtering data - using conditions to select specific rows based on values.