|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |
java.lang.Objectorg.knime.base.node.mine.mds.MDSManager
public class MDSManager
The MDSManager
handling the MDS algorithmic. For each row
of the given DataTable
a DataPoint
with the
specified dimension is created, representing the higher dimensional row.
The DataPoint
s are rearranged in a way that their distances
to each other approximately match the distances of the corresponding high
dimensional points. The rearrangement is an iterative process running as
many epochs as specified. The learn rate, specifying the step size is
reduced after each epoch, so that the process converges at the end.
Field Summary | |
---|---|
static double |
DEF_MINDIST_THRESHOLD
The default value of the minimum distance threshold. |
static int |
DEFAULT_SEED
The default random seed. |
static int |
MAX_SEED
The maximum random seed. |
static int |
MIN_SEED
The minimum random seed. |
Constructor Summary | |
---|---|
MDSManager(int dimension,
String distance,
boolean fuzzy,
BufferedDataTable inData,
ExecutionContext exec)
Creates a new instance of MDSManager with the given
dimension, distance metric, fuzzy flag and in data to use. |
Method Summary | |
---|---|
Hashtable<RowKey,DataPoint> |
getDataPoints()
|
int |
getDimension()
|
void |
init(long seed)
Initializes the lower dimensional data points randomly. |
void |
reset()
Clears the Hashtable containing the high and the
corresponding low dimensional data points. |
void |
train(int epochs,
double learningrate)
Does the training by adjusting the lower dimensional data points accordant to their distances and the distances of the original data. |
Methods inherited from class java.lang.Object |
---|
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
Field Detail |
---|
public static final int DEFAULT_SEED
public static final int MIN_SEED
public static final int MAX_SEED
public static final double DEF_MINDIST_THRESHOLD
Constructor Detail |
---|
public MDSManager(int dimension, String distance, boolean fuzzy, BufferedDataTable inData, ExecutionContext exec) throws IllegalArgumentException
MDSManager
with the given
dimension, distance metric, fuzzy flag and in data to use. If the
dimension is less or equals zero a IllegalArgumentException
is thrown.
dimension
- The output MDS dimensiondistance
- The distance metric to use.fuzzy
- true
if the in data is fuzzy valued data.inData
- The in data to use.exec
- The ExecutionContext
to monitor the
progress.
IllegalArgumentException
- if the specified dimension is less or
equals zero.Method Detail |
---|
public void init(long seed) throws CanceledExecutionException
seed
- The random seed to use.
CanceledExecutionException
- If execution was canceled by the user.public void train(int epochs, double learningrate) throws CanceledExecutionException
epochs
- The number of epochs to train.learningrate
- The learn rate, specifying the step size of
adjustment.
CanceledExecutionException
- If execution was canceled by the user.public Hashtable<RowKey,DataPoint> getDataPoints()
Hashtable
containing the RowKey
s as
as keys and the corresponding lower dimensional DataPoint
s
as values.public void reset()
Hashtable
containing the high and the
corresponding low dimensional data points.
public int getDimension()
|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |