First of all, we must download and then install a Java IDE such as Eclipse, NetBeans and IntelliJ IDEA. I use Eclipse for this tutorial.
Once the IDE of choice is set up, we have to download the MOA framework from its GitHub Repository, then import it into the IDE as a Maven project. In Eclipse, click on File >> Import... >> Maven >> Existing Maven Project. Then, click on Next and Browse... to find unzipped MOA, as follows:
If you encounter a wizard for "Setup Maven plugin connectors", you can overlook it by clicking the Finish button. MOA's packages should be imported as below:
Now, we need to configure settings for running MOA. For that, we need to go to Run >> Run Configurations.... Then, set the Main Class to moa.gui.GUI, or use the Search... button, as below:
Once all is set, you can run MOA by clicking the Run button. Finally, MOA's GUI shows up:
Everything is now ready for getting our hands dirty!
We need to locate a relevant package and put our class in it; it is a simple drag-and-drop. As an example, for my FHDDM drift detector, we should add it under the moa.classifiers.core.driftdetection package.
Once we rerun MOA, we will see the added functionality. For example, in my FHDDM case, we have:
You now have all the nuggets for your experiments. Good luck.
A Few Points
- You may need to update some imports as MOA gets updated over time. For example, you may need to change
import moa.options.IntOption;toimport com.github.javacliparser.IntOption;for FHDDM. - Please contact MOA creators for Java and JRE compatibility, or any warning issues.
- Thank you for your interest in my drift detection methods, and I hope they help you in your research or industrial works.
Ali Pesaranghader © 2019