graphite-dl4j

Perform machine algorithms on graphite query results

License	Apache License Version 2.0
Categories	Ant Build Tools Net Deeplearning4j Business Logic Libraries Machine Learning
GroupId	net.savantly.learning
ArtifactId	graphite-dl4j
Last Version	1.0.0-RELEASE
Release Date	Feb 25, 2018
Type	jar
Description	graphite-dl4j Perform machine algorithms on graphite query results
Project URL	https://github.com/savantly-net/graphite-dl4j
Project Organization	savantly.net
Source Code Management	https://github.com/savantly-net/graphite-dl4j

Download graphite-dl4j

Filename	Size
graphite-dl4j-1.0.0-RELEASE.pom
graphite-dl4j-1.0.0-RELEASE.jar	16 KB
graphite-dl4j-1.0.0-RELEASE-sources.jar	7 KB
graphite-dl4j-1.0.0-RELEASE-javadoc.jar	70 KB

How to add to project

Apache Maven

<!-- https://jarcasting.com/artifacts/net.savantly.learning/graphite-dl4j/ -->
<dependency>
    <groupId>net.savantly.learning</groupId>
    <artifactId>graphite-dl4j</artifactId>
    <version>1.0.0-RELEASE</version>
</dependency>

Gradle Groovy

// https://jarcasting.com/artifacts/net.savantly.learning/graphite-dl4j/
implementation 'net.savantly.learning:graphite-dl4j:1.0.0-RELEASE'

Gradle Kotlin

// https://jarcasting.com/artifacts/net.savantly.learning/graphite-dl4j/
implementation ("net.savantly.learning:graphite-dl4j:1.0.0-RELEASE")

Apache Buildr

'net.savantly.learning:graphite-dl4j:jar:1.0.0-RELEASE'

Apache Ivy

<dependency org="net.savantly.learning" name="graphite-dl4j" rev="1.0.0-RELEASE">
  <artifact name="graphite-dl4j" type="jar" />
</dependency>

Groovy Grape

@Grapes(
@Grab(group='net.savantly.learning', module='graphite-dl4j', version='1.0.0-RELEASE')
)

Scala SBT

libraryDependencies += "net.savantly.learning" % "graphite-dl4j" % "1.0.0-RELEASE"

Leiningen

[net.savantly.learning/graphite-dl4j "1.0.0-RELEASE"]

Dependencies

compile (5)

Group / Artifact	Type	Version
org.slf4j : slf4j-api	jar	1.7.7
com.fasterxml.jackson.core : jackson-databind	jar	2.9.4
org.datavec : datavec-data-codec	jar	0.9.1
org.deeplearning4j : deeplearning4j-core	jar	0.9.1
org.nd4j : nd4j-native-platform	jar	0.9.1

test (4)

Group / Artifact	Type	Version
junit : junit	jar	4.12
ch.qos.logback : logback-classic	jar	1.1.8
org.springframework.boot : spring-boot-starter	jar	1.5.10.RELEASE
org.springframework.boot : spring-boot-starter-test	jar	1.5.10.RELEASE

Project Modules

There are no modules declared in this project.

graphite-dl4j

Perform machine algorithms on graphite query results.

Also see -
graphite client

Usage

<dependency>
	<groupId>net.savantly.learning</groupId>
	<artifactId>graphite-dl4j</artifactId>
	<version>2.0.0-SNAPSHOT</version>
</dependency>

Quickstart

add graphite-dl4j as a dependency to your project.

This example uses a preconfigured multilayer network to train and predict -

private static final String graphiteHost = "127.0.0.1";
private static final int windowSize = 2;
private static final String target = "randomWalk()";
private static final String targetAlias = "random.walk";
private static final int epochs = 20;
private static final int hiddenLayerWidth = 30;
private static final double learningRate = 0.001;

public void testTraining() throws UnknownHostException, SocketException {
	QueryableGraphiteClient client = GraphiteClientFactory.queryableGraphiteClient(graphiteHost);
	PersistentGraphitePredictor learner = 
			PersistentGraphitePredictor.builder(client )
			.setWindowSize(windowSize)
			.setTrainingTarget(target)
			.setTestingTarget(target)
			.setTargetAlias(targetAlias)
			.setEpochs(epochs)
			.setHiddenLayerWidth(hiddenLayerWidth)
			.setLearningRate(learningRate)
			.setRnnLearningRate(learningRate)
			.setUseCache(false) // you can cache/store the model to re-use between app restarts
			.build();
	
	learner.train(); // train the network using the graphite targets
	learner.update(); // make a 2 min prediction and store it in graphite
}

Lower level configuration

	this.graphitePredictor = GraphitePredictor.builder()
			.setClient(client)
			.setEpochs(this.getEpochs())
			.setMiniBatchSize(1)
			.setLearningRate(this.getLearningRate())
			.setRnnLearningRate(this.getRnnLearningRate())
			.setNumOfInputs(windowSize+1)
			.setWindowSize(windowSize)
			.setHiddenLayerWidth(this.getHiddenLayerWidth());

Even lower...

	this.network = RegressionNetwork.builder()
			.setEpochs(epochs)
			.setLearningRate(learningRate)
			.setRnnLearningRate(rnnLearningRate)
			.setTrainingData(trainingData)
			.setNumOfInputs(numOfInputs)
			.setHiddenLayerWidth(hiddenLayerWidth)
			.setIterationListeners(iterationListeners);

Savantly

Savantly custom software solutions for your business

Versions

Version
1.0.0-RELEASE Feb 25, 2018