Skip to content

korriganed/broceliande

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

96 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

broceliande

Keywords: random forest, decision tree, prediction

This project provides a Java implementation of random forests [1, 2]. Random forests use training sets to build decision trees. Given an input (e.g. a person with age, gender, medical background, symptoms) the result (e.g. a disease) of which is unknown, random forests are able to predict the corresponding result.

API

Parameter

The parameters that will be used to build random forests. The default values are :

int minSamplesSplit = 2;
int maxDepth = Integer.MAX_VALUE;
double minImpurityDecrease = 1e-07;
int minSampleLeaf = 1;
int maxFeatures = Integer.MAX_VALUE;
int nbTrees = 10;
Long seed = null;

new Parameter.Builder() : Builder

Return a builder to setup the parameters of the random forest. The available functions to update the default values are :

Builder.minSamplesSplit(int) : Builder
Builder.maxDepth(int) : Builder
Builder.minImpurityDecrease(double) : Builder
Builder.minSampleLeaf(int) : Builder
Builder.maxFeatures(int) : Builder
Builder.seed(Long) : Builder
Builder.nbTrees(int) : Builder
Builder.build() : Parameter
// Builder example
Parameter p = new Parameter.Builder()
                     .nbTrees(200)
                     .maxFeatures(3)
                     .build();

RandomForest

new RandomForest(Parameter) : RandomForest

Constructor of the random forest.

RandomForest.fit(List<D>)

Train the random forest using a list of tuples D.

This function only takes into account the getters of D that are annotated with a Feature which is either ORDERED or CATEGORICAL. The getter of the target (or result) must be annotated by Target with a type which is either CONTINUOUS or DISCRETE.

// Annotation example
@Feature(FeatureType.ORDERED)
  public Integer getAge() {
    return age;
}

@Target(TargetType.DISCRETE)
public Integer getSurvived() {
  return survived;
}

RandomForest.predict(D) : R

Predict the result R according to the data D.

RandomForest.importance() : List<ImmutablePair<String, Double>>

Get the list of features sorted by decreasing importance.

Example

A usage example about Titanic survivors is available at broceliande-example.

References

[1] Leo Breiman. Random Forests. Machine Learning. vol. 45, p. 5-32. 2001.

[2] Gilles Louppe. Understanding random forests: From theory to practice. arXiv preprint arXiv:1407.7502, 2014.

About

random forest, decision tree, prediction

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages