Skip to 0 minutes and 11 secondsWhat is MOA? MOA is open source software that is specifically designed for mining data streams. The most important thing about MOA is that it can handle evolving data streams, data streams that are changing, data streams with concept drift. MOA has many methods – from classification, regression, clustering, frequent pattern mining, outlier detection, concept drift – and is very easy to use and very easy to extend. MOA can be used alone or can be used with Weka or can be used with ADAMS and with MEKA. ADAMS is a very nice workflow engine where you can develop your workflows, and MEKA is a software specific for multi-label learning. MOA runs on a single computer.
Skip to 0 minutes and 51 secondsIn the case that you need to data stream mining in a cluster of computers, like the ones that are used in the Hadoop ecosystem, then Apache SAMOA could be the right tool for you. SAMOA allows you to run stream mining jobs on Apache Storm, Apache S4, Apache Samza, and nowadays Apache Flink. As you know, New Zealand is very famous for birds, so the weka is a native bird of New Zealand. The moa is also a native bird of New Zealand. It’s also flightless as the weka, but nowadays is extinct. As you can see in these pictures, the moa was a very large bird.
Skip to 1 minute and 23 secondsIf you compare it here with the size of a weka or a kiwi, we see that the moa was really large, or here comparing with the size of a human person. Let’s see how to install the MOA package in Weka. We’re going to go to Tools. We’re going to open the Package Manager. From there, we are going to look for the Massive Online Package. Let’s look for it. We should go to the “m”. MassiveOnlineAnalysis. Then we install it. We say “Yes”. OK. And once it’s installed, we go to the Explorer, and then we can see the objects of MOA inside Weka. So there are two types. One is the data generators and the other are the classifiers.
Skip to 2 minutes and 15 secondsFor example, if we went to generate data now, we can look at the MOA generator, and from the MOA generator we can access all the data stream generators inside MOA. OK. Let’s choose one for example. I will generate some data. The other important thing is that we can have access to all the MOA classifiers. So we go inside classifiers inside meta we’ll find this MOA classifier, and from this MOA classifier, we can get access to all the classifiers in MOA. Here are all the classifiers in MOA. Let’s choose, for example, a meta classifier. We can choose the online bagging that’s called OzaBag. And that’s all. In this lesson, we have seen how to install the MOA package in Weka.
Skip to 3 minutes and 5 secondsWe have seen that MOA is open source software specifically designed for data stream mining. It handles evolving data streams, and it has many, many different methods for clustering, classification, regression, frequent pattern mining, outlier detection, and concept drift. It is very easy to use and very easy to extend.
Weka's MOA package
MOA is open source software that is specifically designed for mining data streams. It can handle evolving data streams – ones generated by mechanisms that change, or drift, over time. Like Weka, it contains many data mining methods. Weka’s MOA package gives access to these methods through the Weka Explorer. By the way, the moa was a huge flightless bird that lived in New Zealand until about 500 years ago.
© University of Waikato, New Zealand. CC Creative Commons Attribution 4.0 International License.