Package com.codedstreams.otterstreams.sql.util

Class FeaturePreprocessor

java.lang.Object

com.codedstreams.otterstreams.sql.util.FeaturePreprocessor

public class FeaturePreprocessor
extends Object

Preprocesses features before inference (normalization, encoding, etc.).

Since:

1.0.0

Author:

Nestor Martourez Abiangang A.

Constructor Summary

Constructors

Constructor	Description
FeaturePreprocessor()

Method Summary

Modifier and Type	Method	Description
static Map<String,Object>	fillMissingValues(Map<String,Object> features, Map<String,Object> defaults)	Fills missing values with defaults.
static Map<String,Object>	normalize(Map<String,Object> features, Map<String,Double> minValues, Map<String,Double> maxValues)	Normalizes numerical features to [0, 1] range.
static Map<String,Object>	oneHotEncode(Map<String,Object> features, Map<String,String[]> categoricalMappings)	One-hot encodes categorical features.
static Map<String,Object>	standardize(Map<String,Object> features, Map<String,Double> means, Map<String,Double> stdDevs)	Standardizes features to zero mean and unit variance.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

FeaturePreprocessor

public FeaturePreprocessor()

Method Detail

normalize

public static Map<String,Object> normalize(Map<String,Object> features,
                                                 Map<String,Double> minValues,
                                                 Map<String,Double> maxValues)

Normalizes numerical features to [0, 1] range.

standardize

public static Map<String,Object> standardize(Map<String,Object> features,
                                                   Map<String,Double> means,
                                                   Map<String,Double> stdDevs)

Standardizes features to zero mean and unit variance.

oneHotEncode

public static Map<String,Object> oneHotEncode(Map<String,Object> features,
                                                    Map<String,String[]> categoricalMappings)

One-hot encodes categorical features.

fillMissingValues

public static Map<String,Object> fillMissingValues(Map<String,Object> features,
                                                         Map<String,Object> defaults)

Fills missing values with defaults.