whistleClassifier.training
Class SequentialTrainingSelector
java.lang.Object
whistleClassifier.training.TrainingSelector
whistleClassifier.training.SequentialTrainingSelector
public class SequentialTrainingSelector
- extends TrainingSelector
Selects groups of fragments into training / test groups
simply by concatonating all the files together from some random start point.
- Author:
- Doug Gillespie
|
Method Summary |
int[] |
createSections(int iSpecies,
double trainingFraction,
int sectionLength)
Tell the training selector to create a new (hopefully randomised)
set of training and test data for a species. |
private int[] |
createTrainingSection(TrainingDataGroup trainingDataGroup,
double trainingFraction,
int sectionLength)
|
private int |
fillStores(java.util.Vector<FragmentStore> store,
TrainingDataGroup trainingDataGroup,
int nStores,
int sectionLength,
int iContour)
|
FragmentStore |
getFragmentStore(boolean training,
int iFragmentGroup)
|
(package private) java.util.List<FragmentStore> |
getTestStoreList()
|
(package private) int |
getTotalSections(int iSpecies,
int sectionLength)
Get the total number of training + test sections, each consisting
of sectionLength fragments |
(package private) java.util.List<FragmentStore> |
getTrainingStoreList()
|
| Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
random
private java.util.Random random
trainingFragmentStores
private java.util.Vector<FragmentStore> trainingFragmentStores
testFragmentStores
private java.util.Vector<FragmentStore> testFragmentStores
SequentialTrainingSelector
public SequentialTrainingSelector()
getTotalSections
int getTotalSections(int iSpecies,
int sectionLength)
- Description copied from class:
TrainingSelector
- Get the total number of training + test sections, each consisting
of sectionLength fragments
- Specified by:
getTotalSections in class TrainingSelector
- Parameters:
iSpecies - species index in training setsectionLength - length of each training or test section
- Returns:
- number of complete sections
createSections
public int[] createSections(int iSpecies,
double trainingFraction,
int sectionLength)
- Description copied from class:
TrainingSelector
- Tell the training selector to create a new (hopefully randomised)
set of training and test data for a species.
It does not actually have to
create the data sets at the moment - if it want's to save memory it
can set up appropriate data structures to define what the sets are
and then create them when they are asked for later. Whether or not this
is possible may depend on the type of randomisation and grouping employed
in the concrete class.
- Specified by:
createSections in class TrainingSelector
trainingFraction - fraction of data to be used in training (usually 2/3)
- Returns:
- a two element array giving the number of test and the number of training sections
createTrainingSection
private int[] createTrainingSection(TrainingDataGroup trainingDataGroup,
double trainingFraction,
int sectionLength)
fillStores
private int fillStores(java.util.Vector<FragmentStore> store,
TrainingDataGroup trainingDataGroup,
int nStores,
int sectionLength,
int iContour)
getFragmentStore
public FragmentStore getFragmentStore(boolean training,
int iFragmentGroup)
- Specified by:
getFragmentStore in class TrainingSelector
getTrainingStoreList
java.util.List<FragmentStore> getTrainingStoreList()
- Specified by:
getTrainingStoreList in class TrainingSelector
getTestStoreList
java.util.List<FragmentStore> getTestStoreList()
- Specified by:
getTestStoreList in class TrainingSelector