[Python] Retrieving and Updating Row Data in Pandas DataFrames (loc / iloc)

2026年1月4日

In Pandas DataFrame operations, extracting data row by row is just as frequent as retrieving column data. When you need to extract records with specific IDs or access data by its n-th position, you use the loc and iloc properties.

This article explains the differences between these two accessors, how to use them, and how to update row data.

Differences Between loc and iloc

There are two main ways to specify row data in Pandas:

loc["label"]: Access by specifying the index name (row label).
iloc[integer]: Access by specifying the row number (integer starting from 0).

Implementation Sample Code

Below is code that creates a DataFrame to manage student test scores, then retrieves and updates data for a specific student.

import pandas as pd

def manage_row_data():
    """
    Function demonstrating retrieval and updating of row data using loc and iloc
    """
    
    # 1. Preparation of Data
    # DataFrame with student names as index, and Math/English scores as columns
    student_names = ["Alice", "Bob", "Charlie", "David", "Ellen"]
    exam_data = {
        "Math": [75, 82, 90, 68, 95],
        "English": [88, 79, 85, 92, 70]
    }
    
    # Create DataFrame specifying index by name
    df = pd.DataFrame(exam_data, index=student_names)
    
    print("--- Initial Exam Data ---")
    print(df)
    print("\n")


    # 2. Retrieving row using loc (Label Specification)
    print("=== Row Retrieval via loc (Label) ===")
    # Get row data for "Charlie"
    # Return value is a Series
    charlie_data = df.loc["Charlie"]
    
    print("--- Charlie's Data ---")
    print(charlie_data)
    print(f"Type: {type(charlie_data)}\n")


    # 3. Retrieving row using iloc (Position Specification)
    print("=== Row Retrieval via iloc (Position) ===")
    # Get the 4th row from the top (Index 3) -> David
    row_at_3 = df.iloc[3]
    
    print("--- 4th Data (David) ---")
    print(row_at_3)
    
    # Extract value of specific column from retrieved Series
    # Dot notation or bracket notation can be used
    david_math = row_at_3.Math
    print(f">> David's Math Score: {david_math}\n")


    # 4. Updating Row Data
    print("=== Updating Row Data ===")
    
    # Create a new Series to correct Alice's scores
    # Index (column names) must match the DataFrame
    new_alice_scores = pd.Series([80, 95], index=["Math", "English"])
    
    # Overwrite Alice's entire row using loc
    df.loc["Alice"] = new_alice_scores
    
    print("--- Data After Update (Alice's scores changed) ---")
    print(df)

if __name__ == "__main__":
    manage_row_data()

Execution Result

--- Initial Exam Data ---
         Math  English
Alice      75       88
Bob        82       79
Charlie    90       85
David      68       92
Ellen      95       70


=== Row Retrieval via loc (Label) ===
--- Charlie's Data ---
Math       90
English    85
Name: Charlie, dtype: int64
Type: <class 'pandas.core.series.Series'>

=== Row Retrieval via iloc (Position) ===
--- 4th Data (David) ---
Math       68
English    92
Name: David, dtype: int64
>> David's Math Score: 68

=== Updating Row Data ===
--- Data After Update (Alice's scores changed) ---
         Math  English
Alice      80       95
Bob        82       79
Charlie    90       85
David      68       92
Ellen      95       70

Explanation

Return Value is a Series

When you retrieve only one row using loc or iloc, the return value becomes a Series object where the column names serve as the index. Therefore, you can easily access individual values using syntax like row_data.ColumnName.

Updating Rows

When updating a row, the shape of the new data (list or Series) must match the column structure of the DataFrame. While it is possible to assign a list directly like df.loc["Label"] = [Val1, Val2], using pd.Series clarifies the correspondence of labels (column names) and allows for safer updates.

よかったらシェアしてね！