Example Project for using ELKI

This project uses Maven to download ELKI.

The Eclipse m2e extensions should work fine with this project. If necessary, convert to a Maven project, and run Maven Update.

Building a package:

The Maven package goal will produce a package in the output folder target.

All dependencies will be copied into the subdirectory dependency.

To launch this, use mvn package on the command line, or in Eclipse Run As, Maven Build and specify the goal package.

Custom Algorithm Example:

This example algorithm computes the mean and variance of each attribute.

Launching using Maven:

From command line, you can launch using mvn -P launch test.

Launching from the jar package:

First, build a package as explained above via mvn package.

java -jar yourpackage.jar will spawn the MiniGUI, where you can choose your algorithm in the -algorithm parameter dropdown.

Launching in Eclipse:

To launch use the included Eclipse launch configuration "ELKI Example".

Custom Application Example:

This application runs above algorithm multiple times.

Launching from the jar package:

First, build a package as explained above via mvn package.

To launch the command line version (great for automated experiments):

java -jar yourjar.jar example.ExampleApplication

To launch the MiniGUI version, use:

java -jar yourjar.jar minigui example.ExampleApplication

As you can see, ELKI added an option to choose the number of repetitions, the new parameter that we have added in our application.

Known limitations:

As of ELKI 0.7.0, classes in the default package cause problems in the MiniGUI. Therefore, always use a package for your classes. This will be fixed in the next ELKI version.

To build and distribute an add-on for ELKI, set the addClasspath function of the maven-jar-plugin to false to disable adding dependencies to the jar manifest.