PreTab

PreTab is a modular, scikit-learn-compatible representation and preprocessing library for tabular data. Its focus is feature representation and expansion: splines, neural basis maps, piecewise-linear encoding, kernel approximations, and language embeddings, each shipped as a standalone transformer that speaks the standard fit / transform API. Every PreTab transformer subclasses BaseEstimator and TransformerMixin, so it drops into a Pipeline or ColumnTransformer alongside any sklearn-native transformer you already use.

Note

These docs are for PreTab 1.0.0.

Highlights

🔢 Numerical preprocessing

Spline expansions (B-splines, natural cubic, thin-plate, tensor-product, P-splines), neural basis maps (RBF, ReLU, sigmoid, tanh), custom binning, and Piecewise Linear Encoding (PLE).

🌤 Categorical preprocessing

Ordinal and one-hot encodings, plus pretrained language embeddings for high-cardinality or semantic columns.

🔧 Composable pipelines

Fully compatible with sklearn.pipeline.Pipeline and sklearn.compose.ColumnTransformer; compose standalone representations with scikit-learn transformers and estimators.

🧠 Smart defaults

Automatic feature-type detection (numerical vs. categorical) with support for both pandas.DataFrame and numpy.ndarray inputs.

See it in action

import numpy as np
import pandas as pd
from pretab import Preprocessor

df = pd.DataFrame({
    "age": np.random.randint(18, 65, size=100),
    "income": np.random.normal(60_000, 15_000, size=100).astype(int),
    "city": np.random.choice(["Berlin", "Munich", "Hamburg"], size=100),
})
y = np.random.randn(100)

# One strategy per feature type: PLE for numerics, integer codes for categoricals
pre = Preprocessor(numerical_method="ple", categorical_method="int")
X = pre.fit_transform(df, y)          # single stacked, model-ready array
X.shape
# (100, 15)

Preprocessor detects the column types, fits a strategy per column, and returns a single stacked, model-ready array by default, so it drops straight into a plain sklearn.pipeline.Pipeline; pass output_structure="blocks" for a dict of per-feature blocks instead. Inspect the resolved layout at any time with get_feature_info(verbose=True):

feature  kind         pipeline                        dim   cats
----------------------------------------------------------------
age      numerical    imputer -> minmax -> ple          7      -
income   numerical    imputer -> minmax -> ple          7      -
city     categorical  imputer -> continuous_ordinal     1      4

Mix strategies per column

Columns rarely want the same treatment. Pass a feature_preprocessing map to give each column its own strategy, and a single fit still returns one coherent feature set:

pre = Preprocessor(feature_preprocessing={
    "age": "ple",       # piecewise-linear encoding
    "income": "rbf",    # radial-basis expansion
    "city": "one-hot",  # one-hot categorical
})
X = pre.fit_transform(df, y)

X.shape
# (100, 17)