In Polars, you can change the position of a column in a DataFrame using the select() method by reordering the columns explicitly. In this article, I will explain how to move the last column to the first, shift the first column to the end, or reposition a middle column to either the beginning or the end with examples.
Key Points–
- The primary method to rearrange columns is
df.select(), where you specify the new column order as a list. - The
select()method allows you to reorder columns by specifying the desired order manually. - Extract the target column and prepend it to the list of remaining columns.
- You can rearrange multiple columns simultaneously by specifying a new order in
select(). - Use regex or condition-based filtering to select and reposition columns dynamically.
- Get the last column dynamically using
df.columns[-1]and reorder accordingly. pl.all()selects all columns; combine withexclude()to omit specific columns.- Create a function that moves a column to the start or end for flexibility.
<a id="h-usage-of-change-position-of-a-column-in-polars"></a>
Usage of Change Position of a Column in Polars
You can change the order of columns in a Polars DataFrame using the select() method. Simply specify the columns in the desired order.
To run some examples of how to change position of a column in polars, let’s create a Polars DataFrame.
import polars as pl
technologies= {
'Courses':["Spark","PySpark","Hadoop", "Python"],
'Fee' :[22000,25000,23000, 24000],
'Duration':['30days','50days','60days', '45days'],
'Discount':[1000,2300,2000, 1200]
}
df = pl.DataFrame(technologies)
print("Original DataFrame:\n", df)
Yields below output.
To reorder columns in a Polars DataFrame using select() method, you can specify the desired column order.
# Reorder columns: Discount first, then Courses, Fee, Duration
df2 = df.select(["Discount", "Courses", "Fee", "Duration"])
print("Reordered DataFrame:\n", df2)
Here,
select()is used to choose and reorder columns.- The order of columns in the output matches the order in the list passed to
select(). - In this example, we changed the order from
["Courses", "Fee", "Duration", "Discount"]to["Discount", "Courses", "Fee", "Duration"].
<a id="h-move-single-column-to-front"></a>
Move Single Column to Front
To move a single column to the front (first position) in a Polars DataFrame, you can use select() along with pl.col() and pl.all() functions.
# Move "Discount" to the front
df2 = df.select(["Discount"] + [col for col in df.columns if col != "Discount"])
print("Reordered DataFrame (Courses moved to last):\n", df2)
# Move "Duration" to the front
df2 = df.select(
pl.col("Duration"),
pl.all().exclude("Duration"))
print("Reordered DataFrame (Courses moved to last):\n", df2)
# Output:
# Reordered DataFrame (Courses moved to last):
# shape: (4, 4)
┌──────────┬─────────┬───────┬──────────┐
│ Discount ┆ Courses ┆ Fee ┆ Duration │
│ --- ┆ --- ┆ --- ┆ --- │
│ i64 ┆ str ┆ i64 ┆ str │
╞══════════╪═════════╪═══════╪══════════╡
│ 1000 ┆ Spark ┆ 22000 ┆ 30days │
│ 2300 ┆ PySpark ┆ 25000 ┆ 50days │
│ 2000 ┆ Hadoop ┆ 23000 ┆ 60days │
│ 1200 ┆ Python ┆ 24000 ┆ 45days │
└──────────┴─────────┴───────┴──────────┘
Here,
pl.all()selects all columns.pl.all().exclude("col_name")selects all columns except the specified one.select()is flexible you can reorder columns in any way you like.
<a id="h-move-column-to-specific-position"></a>
Move Column to Specific Position
To move a column to a specific position (not just the front) in a Polars DataFrame, you can use select() with a custom column order.
def move_column(df: pl.DataFrame, col: str, pos: int) -> pl.DataFrame:
cols = df.columns
if col not in cols:
raise ValueError(f"Column '{col}' not found!")
new_order = cols[:pos] + [col] + [c for c in cols if c != col and c not in cols[:pos]]
return df.select(new_order)
# Move "Duration" to index 0 (first position)
df2 = move_column(df, "Duration", 0)
print("After changing the position of the columns:\n", df2)
# Output:
# After changing the position of the columns:
# shape: (4, 4)
┌──────────┬─────────┬───────┬──────────┐
│ Duration ┆ Courses ┆ Fee ┆ Discount │
│ --- ┆ --- ┆ --- ┆ --- │
│ str ┆ str ┆ i64 ┆ i64 │
╞══════════╪═════════╪═══════╪══════════╡
│ 30days ┆ Spark ┆ 22000 ┆ 1000 │
│ 50days ┆ PySpark ┆ 25000 ┆ 2300 │
│ 60days ┆ Hadoop ┆ 23000 ┆ 2000 │
│ 45days ┆ Python ┆ 24000 ┆ 1200 │
└──────────┴─────────┴───────┴──────────┘
<a id="h-move-last-column-to-first-position"></a>
Move Last Column to First Position
You can move a column from the last position to the first in Polars using the select() method. Now let’s see how to move the last column to the first position in Polars DataFrame.
# Move 'Discount' (last column) to the first position
columns = df.columns
new_order = [columns[-1]] + columns[:-1]
df2 = df.select(new_order)
print("Reordered DataFrame:\n", df2)
# Output:
# Reordered DataFrame:
# shape: (4, 4)
┌──────────┬─────────┬───────┬──────────┐
│ Discount ┆ Courses ┆ Fee ┆ Duration │
│ --- ┆ --- ┆ --- ┆ --- │
│ i64 ┆ str ┆ i64 ┆ str │
╞══════════╪═════════╪═══════╪══════════╡
│ 1000 ┆ Spark ┆ 22000 ┆ 30days │
│ 2300 ┆ PySpark ┆ 25000 ┆ 50days │
│ 2000 ┆ Hadoop ┆ 23000 ┆ 60days │
│ 1200 ┆ Python ┆ 24000 ┆ 45days │
└──────────┴─────────┴───────┴──────────┘
Here,
- Get column names:
df.columns - Rearranging columns:
columns[-1]: Retrieves the last column (Discount).columns[:-1]: Retrieves all columns except the last.new_order = [columns[-1]] + columns[:-1]: Moves the last column to the first position.
- Apply new order:
df.select(new_order)
<a id="h-move-multiple-columns-by-pattern"></a>
Move Multiple Columns by Pattern
You can rearrange multiple columns based on a pattern dynamically. This is useful when you need to group specific columns together or reorder them based on a naming convention.
# Define a pattern-based column selection
pattern = ["Fee", "Discount"] # Move these to the front
# Separate columns matching the pattern and the rest
matching_cols = [col for col in df.columns if any(p in col for p in pattern)]
remaining_cols = [col for col in df.columns if col not in matching_cols]
# New column order
new_order = matching_cols + remaining_cols
# Reorder DataFrame
df2 = df.select(new_order)
print("Reordered DataFrame Matching columns moved to the front:\n", df2)
# Output:
# Reordered DataFrame Matching columns moved to the front:
shape: (4, 4)
┌───────┬──────────┬─────────┬──────────┐
│ Fee ┆ Discount ┆ Courses ┆ Duration │
│ --- ┆ --- ┆ --- ┆ --- │
│ i64 ┆ i64 ┆ str ┆ str │
╞═══════╪══════════╪═════════╪══════════╡
│ 22000 ┆ 1000 ┆ Spark ┆ 30days │
│ 25000 ┆ 2300 ┆ PySpark ┆ 50days │
│ 23000 ┆ 2000 ┆ Hadoop ┆ 60days │
│ 24000 ┆ 1200 ┆ Python ┆ 45days │
└───────┴──────────┴─────────┴──────────┘
<a id="h-conclusion"></a>
Conclusion
In conclusion, changing the position of columns in Polars is a powerful yet simple operation that improves data readability, consistency, and processing efficiency. By using the select() method, you can dynamically reposition single or multiple columns to the beginning, end, or any specific location. This is especially beneficial for tasks such as data preprocessing, reporting, merging, exporting, and visualization.
Happy Learning !!
<a id="h-related-articles"></a>
Related Articles
- Convert Polars String to Integer
- Polars Sum Multiple Columns
- How to Drop Row in Polars
- Select Polars Columns by Index
- Convert Polars Cast Integer to Float
- Convert Polars Cast Float to Integer
- Polars Filter by Column Value
- Polars Sum DataFrame Columns With Examples
- Polars DataFrame shape – Explained by Examples
- Reorder Columns in a Specific Order Using Polars
<a id="h-references"></a>