Feature Engineering with Polars

Feature engineering done with Polars

How to install

pip install feature-engineering-polars

How to use it

import polars as pl
from fe_polars.imputing.base_imputing import Imputer
from fe_polars.encoding.target_encoding import TargetEncoder

dataframe = pl.DataFrame(
        {
            "City": ["A", "A", "B", "B", "B", "C", "C", "C"],
            "Rain": [103, None, 90, 75, None, 200, 155, 127],
            "Temperature": [30.5, 32, 25, 38, 40, 29.6, 21.3, 24.9],
        }
    )

imputer = Imputer(features_to_impute=["Rain"], strategy="mean")
encoder = TargetEncoder(smoothing=2, features_to_encode=["City"])

temp = imputer.fit_transform(x=dataframe)
encoder.fit_transform(x=temp, y=dataframe['Temperature'])


shape: (8, 3)
City    Temperature Rain
f64     f64         f64

30.706  30.5        103.0
30.706  32.0        125.0
32.665  25.0        90.0
32.665  38.0        75.0
32.665  40.0        125.0
27.225  29.6        200.0
27.225  21.3        155.0
27.225  24.9        127.0

Available transformers

• Encoding:
  • Target encoding
  • One hot encoding
• Imputing:
  • Base imputing:
    • Mean imputing
    • Median imputing
    • Max imputing
    • Min imputing
    • Fixed value imputing