Implements the isolation forest method for anomaly detection
The data is expected to have two class values for the class attribute, which is ignored at training time.The distributionForInstance() method returns the anomaly score as the first element in the distribution, the second element is one minus this score.
To evaluate performance of this method for a dataset where anomalies are known, simply code the anomalies using the class attribute: normal cases should correspond to the second value of the class attribute, anomalies to the first one.
For more information, see:
Fei Tony Liu, Kai Ming Ting, Zhi-Hua Zhou: Isolation Forest.
In: ICDM, 413-422, 2008.
(based on WEKA 3.7)
For further options, click the 'More' - button in the dialog.
All weka dialogs have a panel where you can specify classifier-specific parameters.