Expand description
Data preprocessing transformers.
Provides scalers, encoders, dimensionality reduction, and a
composable Transformer trait for building preprocessing pipelines.
Structs§
- Column
Transformer - Apply different transformers to different column subsets, then concatenate all transformed outputs.
- Label
Encoder - Encode string labels as integer indices.
- MinMax
Scaler - Scale features to a [0, 1] range.
- Normalizer
- Normalize samples individually to unit norm.
- OneHot
Encoder - One-hot encoder for integer-encoded categorical features.
- Pca
- Principal Component Analysis.
- Polynomial
Features - Generate polynomial and interaction features.
- Robust
Scaler - Scale features using the median and inter-quartile range (IQR).
- Simple
Imputer - Imputes missing (
NaN) values in each feature column. - Standard
Scaler - Standardize features by removing the mean and scaling to unit variance.
Enums§
- Drop
Strategy - Strategy for dropping one-hot columns to avoid multicollinearity.
- Norm
- Norm type for row-wise normalization.
- Strategy
- Strategy for computing the replacement value per feature.
- Unknown
Strategy - Strategy for handling categories seen at transform time but not at fit time.
Traits§
- Transformer
- A data transformer that can be fitted on a dataset and applied to transform it.