5 Ways to Delete a Column in Pandas DataFrame

5 Ways to Delete a Column in Pandas DataFrame

Cleaning a dataset often means removing columns that are redundant, duplicated, or just not useful for the analysis at hand. Pandas—the go-to Python library for tabular data—offers several ways to drop columns. Below are five practical approaches, complete with runnable examples and tips on when to reach for each one.

For the snippets below, start with the same sample DataFrame so you can see how each method behaves:

import pandas as pd
 
# Sample data
df = pd.DataFrame(
    {
        "name": ["Ada", "Bob", "Chloe"],
        "age": [29, 35, 42],
        "city": ["London", "Berlin", "New York"],
        "score": [88, 91, 77],
    }
)
 
print(df)

1. Using drop()

drop() is the most flexible option. You can remove a single column or multiple columns, choose whether to modify the original DataFrame, and even drop rows by switching the axis.

# Drop a single column without touching the original
df_without_city = df.drop(columns=["city"])
print(df_without_city)
 
# Drop multiple columns in place
df.drop(columns=["city", "score"], inplace=True)
print(df)

• Pass columns=[...] or axis=1 to make it clear you are dropping columns.
• Use inplace=True when you no longer need the original data. Otherwise, omit it so you keep an unchanged copy.

2. Using the del statement

Python’s built-in del removes a column immediately from the DataFrame—no copy is created.

del df["score"]
print(df)

• Pros: Short and expressive.
• Cons: The deletion is permanent and raises a KeyError if the column name is wrong. Reach for it when you are scripting quick cleanups and are confident about your column names.

3. Using pop()

pop() behaves like the list method of the same name: it removes the column and returns the data as a Series. This is perfect when you want to reuse the values elsewhere.

scores = df.pop("score")
print(df)
print(scores)

• After calling pop(), the column is gone from df.
• The returned Series keeps the original index so you can align it with other data structures or store it for later.

4. Column filtering

Instead of deleting columns, you can rebuild the DataFrame with only the fields you need. This technique avoids modifying the original object and makes the selected columns explicit.

subset = df[["name", "score"]]
print(subset)

• Handy when you need a lightweight projection of the data for plotting, serialization, or quick debugging.
• Combine with .copy() if you want to ensure the subset is fully independent of the original DataFrame.

5. Using loc or iloc

loc and iloc allow you to filter columns by label or position. They are especially useful when you want to drop columns based on a condition.

# Keep every column except "score" using loc
no_score = df.loc[:, df.columns != "score"]
print(no_score)
 
# Keep the first and last columns by index with iloc
ends_only = df.iloc[:, [0, -1]]
print(ends_only)

• loc works with column names, while iloc uses zero-based integer positions.
• Combine column masks with string methods (df.columns.str.contains(...)) to exclude patterns in one line.

Summary table

Conclusion

Pick the deletion method that matches your workflow:

• Reach for drop() when you need versatility or want to remove several columns at once.
• Use del for straightforward, in-place cleanups inside scripts.
• Prefer pop() when you plan to reuse the removed data.
• Lean on column filtering or loc/iloc whenever you are shaping a view of the data rather than mutating it.

For large DataFrames, avoid unnecessary copies—prefer in-place operations when you no longer need the original column. With these five patterns, you can keep your pandas pipelines tidy and focused on the columns that matter.