- 1: Pure default feature extraction using Batch.
- 2:Feature extraction using custom features and aggregators using Batch
Using the Batch class with default settings
- Create a Batch
setFetaures(RecordingInfo files) to load non-file based samples.
Using the Batch Class with custom settings
- setFeatures(HashMap<String,Boolean>, HashMap<String,String> toggles) - establishes which features to extract and with what parameters.
- applySettings(RecordingInfo recording, int windowSize, double windowOverlap, double samplingRate, boolean normalise, boolean perWindow, boolean overall, java.lang.String destinationFK, java.lang.String destinationFV, int outputType)
The recording list is to pass the array by reference and should be of length 1. The remainder of the entries are as described in the GUI description.
- setAggregators(java.lang.String aggNames, java.lang.String aggFeatures, java.lang.String aggParam):
The aggNames are the names listed in the name descriptor of the AggregatorDefinition. The aggFeatures parameter is an array
where every specific aggregator's feature list is listed in the last dimension of the array listing (in order) each feature
by name in its FeatureDefinition. Similarly, aggregators requiring parameters set them in the string array located in the
same index in the aggParam array.
- (Optional:) utilize DataModel.getDataModel() and set the public output streams variables .featureKey and .featureValue to output to a
stream rather than a file.
The names of aggregators and features are easiest extracted from the source, but can be obtained from their
metadata (FeatureDefinition or AggregatorDefinition) available from each aggregator and feature.
The set of all features is available from the DataModel in the
aggregatorMap public variables.
To retrieve the results of a computation without using XML or Weka ARFF, use .getDataModel().container.getResults(); which returns a double containing all results.