Dynamically index a column in Polars

Question

I have a simple dataframe look like this:

import polars as pl

df = pl.DataFrame({ 
    'ref': ['a', 'b', 'c', 'd', 'e', 'f'], 
    'idx': [4, 3, 1, 6, 2, 5], 
})

How can I obtain the result as creating a new column as ref[idx], which is dynamic index from another column?

out = pl.DataFrame({     
    'ref': ['a', 'b', 'c', 'd', 'e', 'f'],     
    'idx': [4, 3, 1, 6, 2, 5],     
    'ref[idx]': ['d', 'c', 'a', 'f', 'b', 'e'], 
})

shape: (6, 3)
┌─────┬─────┬──────────┐
│ ref ┆ idx ┆ ref[idx] │
│ --- ┆ --- ┆ ---      │
│ str ┆ i64 ┆ str      │
╞═════╪═════╪══════════╡
│ a   ┆ 4   ┆ d        │
│ b   ┆ 3   ┆ c        │
│ c   ┆ 1   ┆ a        │
│ d   ┆ 6   ┆ f        │
│ e   ┆ 2   ┆ b        │
│ f   ┆ 5   ┆ e        │
└─────┴─────┴──────────┘

jqurious · Accepted Answer · 2025-09-28 08:10:42Z

4

Polars has .get() / .gather() expressions for extracting values by index.

df.with_columns(
    pl.col("ref").get(pl.col("idx") - 1).alias("ref[idx]")
)

shape: (6, 3)
┌─────┬─────┬──────────┐
│ ref ┆ idx ┆ ref[idx] │
│ --- ┆ --- ┆ ---      │
│ str ┆ i64 ┆ str      │
╞═════╪═════╪══════════╡
│ a   ┆ 4   ┆ d        │
│ b   ┆ 3   ┆ c        │
│ c   ┆ 1   ┆ a        │
│ d   ┆ 6   ┆ f        │
│ e   ┆ 2   ┆ b        │
│ f   ┆ 5   ┆ e        │
└─────┴─────┴──────────┘

answered Sep 28 at 8:10

jqurious

24.2k6 gold badges24 silver badges43 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

wim · Accepted Answer · 2025-09-27 22:40:01Z

1

The DataFrame.with_columns() method could be used to add another column. Subtract one from idx for 0-based indexing.

>>> df.with_columns(**{'ref[idx]': df['ref'][df['idx']-1]})
shape: (6, 3)
┌─────┬─────┬──────────┐
│ ref ┆ idx ┆ ref[idx] │
│ --- ┆ --- ┆ ---      │
│ str ┆ i64 ┆ str      │
╞═════╪═════╪══════════╡
│ a   ┆ 4   ┆ d        │
│ b   ┆ 3   ┆ c        │
│ c   ┆ 1   ┆ a        │
│ d   ┆ 6   ┆ f        │
│ e   ┆ 2   ┆ b        │
│ f   ┆ 5   ┆ e        │
└─────┴─────┴──────────┘

answered Sep 27 at 22:40

wim

368k114 gold badges681 silver badges817 bronze badges

3 Comments

Baffin Chu Sep 27 at 23:53

Thank you very much for the prompt response, may I also ask how can u format the dataframe like the one in your answer?

wim Sep 28 at 6:27

That’s just how they pretty print by default in IPython.

etrotta Sep 28 at 18:25

Note that df["col"] series operations will not work in Lazy mode

etrotta · Accepted Answer · 2025-09-28 18:23:36Z

1

You could use a join,

right = df.select(pl.row_index("index")+1, pl.col("ref").alias("ref[index]"))

df.join(right, left_on="idx", right_on="index")

answered Sep 28 at 18:23

etrotta

1,0701 silver badge9 bronze badges

Collectives™ on Stack Overflow

Dynamically index a column in Polars

3 Answers 3

Comments

3 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related