CODEWITHSUNDEEP
pythonpandasgroup-bypython-polars
jbssm
jbssm

Reputation: 7161

How to use group_by and apply a custom function with Polars?

I am breaking my head trying to figure out how to use group_by and apply a custom function using Polars.

Coming from Pandas, I was using:

import pandas as pd
from scipy.stats import spearmanr
 
def get_score(df):
   return spearmanr(df["prediction"], df["target"]).correlation

df = pd.DataFrame({
    "era": [1, 1, 1, 2, 2, 2, 5],
    "prediction": [2, 4, 5, 190, 1, 4, 1],
    "target": [1, 3, 2, 1, 43, 3, 1]
})

correlations = df.groupby("era").apply(get_score)

Polars has map_groups() to apply a custom function over groups, which I tried:

correlations = df.group_by("era").map_groups(get_score)

But this fails with the error message:

'Could not get DataFrame attribute '_df'. Make sure that you return a DataFrame object.: PyErr { type: <class 'AttributeError'>, value: AttributeError("'float' object has no attribute '_df'"), traceback: None }

Any ideas?

Upvotes: 22

Views: 21627

Answers (1)

ritchie46
ritchie46

Reputation: 14710

Polars has the pl.corr() function which supports method="spearman"

If you want to use a custom function you could do it like this:

Custom function on multiple columns/expressions

import polars as pl
from typing import List
from scipy import stats

df = pl.DataFrame({
    "g": [1, 1, 1, 2, 2, 2, 5],
    "a": [2, 4, 5, 190, 1, 4, 1],
    "b": [1, 3, 2, 1, 43, 3, 1]
})

def get_score(args: List[pl.Series]) -> pl.Series:
    return pl.Series([stats.spearmanr(args[0], args[1]).correlation], dtype=pl.Float64)

(df.group_by("g", maintain_order=True)
 .agg(
    pl.map_groups(
        exprs=["a", "b"], 
        function=get_score).alias("corr")
 ))

Polars provided function

(df.group_by("g", maintain_order=True)
 .agg(
     pl.corr("a", "b", method="spearman").alias("corr")
 ))

Both output:

shape: (3, 2)
┌─────┬──────┐
│ g   ┆ corr │
│ --- ┆ ---  │
│ i64 ┆ f64  │
╞═════╪══════╡
│ 1   ┆ 0.5  │
│ 2   ┆ -1.0 │
│ 5   ┆ NaN  │
└─────┴──────┘

Custom function on a a single column/expression

We can also apply custom functions on single expressions, via .map_elements

Below is an example of how we can square a column with a custom function and with normal polars expressions. The expression syntax should always be preferred, as its a lot faster.

(df.group_by("g")
 .agg(
     pl.col("a").map_elements(lambda group: group**2).alias("squared1"),
     (pl.col("a")**2).alias("squared2")
 ))

Upvotes: 34

Related Questions