Package autoweka

Interface Summary
ClassifierResult.Metric Generic interface for different Metrics You can define custom metrics and use them by using their full class name
 

Class Summary
ApplicabilityTester Utility class for determining what kind of methods can be applied to a set of instances.
ApplicabilityTester.ApplicableClassifiers  
ClassifierResult Wraps all the data that comes from training a classifier so it can be passed back to a SMBO method
ClassifierResult.AreaAboveROC  
ClassifierResult.ErrorRateMetric  
ClassifierResult.MeanAbsoluteErrorMetric  
ClassifierResult.RelativeAbsoluteErrorMetric  
ClassifierResult.RootMeanSquaredErrorMetric  
ClassifierResult.RootRelativeSquaredErrorMetric  
ClassifierRunner Class that is responsible for actually running a WEKA classifier from start to finish using the Auto-WEKA argument format.
ClassifierRunner.EvaluatorThread  
ClassParams Captures all the Parameters and Conditionals for a specific class that Auto-WEKA will search over
Conditional Simple struct that matains conditionality between params
DefaultsExperimentConstructor Builds a list experiment that only tries to run the defaults for all valid classifiers
Experiment  
Experiment.TrajectoryPointExtra Class capturing any of the extra dataset partitions that should be run on all the trajectory points for analysis
ExperimentBatch Helpful XML class that easily allows for the creation of multiple experiments.
ExperimentBatch.DatasetComponent Enscapsulates a dataset that you want to run experiments on
ExperimentBatch.ExperimentComponent Captures all the settings of an experiment that are not related to a dataset
ExperimentConstructor Abstract class responsible for generating all the necessary stuff to run an Auto-WEKA Experiment.
GridSearchExperimentConstructor Creates a ListExperiment that does a grid search over the allowable classifiers
InstanceGenerator Abstract class for partitioning a dataset (consisting of training and testing data) into 'instances' that are given to the SMBO method
InstanceGenerator.NestedArgs Struct for dealing with nested args on InstanceGenerators that are chained together.
ListExperimentConstructor  
Parameter Captures all the data about if this parameter is categorical, numeric (and what the ranges are), and some other helper functions to get some data out of them
ParameterConditionalGroup  
SubProcessWrapper An implementation of a Wrapper that should ideally just be called from other Java objects - it performes the training/evaluation of a classifier on a sub process to preserve the JVM of the caller
SubProcessWrapper.ErrorAndTime Small class that just encapsulates the error and time of a sub process run TODO: Should really be changed to use a ClassifierResult instead
Trajectory Class that captures a trajectory of an optimisation run
Trajectory.Point Representation of a single point along the optimisation run
Trajectory.Point.SpecificInstanceInfo  
TrajectoryGroup Wrapper class that provides a nice way to get at a bunch of trajectories from an experiment run with different seeds
TrajectoryMerger Given a bunch of folders that have experiments in them, merges all the trajectories into a single file for easy analysis
TrajectoryParser Abstract class that SMBO Methods must provide that allows for trajectories to be extracted after a run has completed
TrajectoryPointExtraRunner Helper class that runs all the trajectoryPointExtras on a given trajectory - needed for analysis experiments
TrajectoryPointPredictionRunner  
Util Bunches of random stuff that seems to be handy in all sorts of places
Util.NullPrintStream Utility to class to make a void print stream
Util.ProcessKillerShutdownHook  
WekaArgumentConverter Utility class that can convert arguments from Auto-WEKA to WEKA
WekaArgumentConverter.Arguments  
Wrapper Generic class that gets called from an SMBO method, completes the evaluation, and returns the result back up to the SMBO method.
XmlSerializable Helper class to make objects easily be read to/from XML, as well as specifying them on the command line
 

Enum Summary
Parameter.ParamType