org.knime.base.node.preproc.sample
Class Sampler

java.lang.Object
  org.knime.base.node.preproc.sample.Sampler

public final class Sampler
extends Object

Utility class that allows to create row filters for sampling.

Author:

Bernd Wiswedel, University of Konstanz

Method Summary

static RowFilter	createRangeFilter(DataTable table, double fraction, ExecutionMonitor exec) Creates a filter that to filter the first 100 * fraction rows from a table.
static RowFilter	createRangeFilter(int count) Creates a filter that passes only the first count rows.
static RowFilter	createSampleFilter(DataTable table, double fraction, ExecutionMonitor exec) Creates row filter that samples precisely a given fraction of rows.
static RowFilter	createSampleFilter(DataTable table, double fraction, Random rand, ExecutionMonitor exec) Creates row filter that samples precisely a given fraction of rows.
static RowFilter	createSampleFilter(DataTable table, int count, ExecutionMonitor exec) Creates row filter that samples arbitrary count rows from table.
static RowFilter	createSampleFilter(DataTable table, int count, Random rand, ExecutionMonitor exec) Creates row filter that samples arbitrary count rows from table.
static RowFilter	createSampleFilter(double fraction) Creates row filter that randomly samples about 100 * fraction percent from a table.
static DataTable	createSamplingTable(DataTable table, RowFilter filter) Convenience method that creates a new DataTable that samples rows according to a given row filter.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

createSamplingTable

public static final DataTable createSamplingTable(DataTable table,
                                                  RowFilter filter)

Convenience method that creates a new DataTable that samples rows according to a given row filter.

Parameters:

table - the table to wrap, i.e. to sample from

filter - the filter to use

Returns:

a new RowFilterTable

See Also:

RowFilterTable.RowFilterTable(DataTable, RowFilter)

createRangeFilter

public static final RowFilter createRangeFilter(DataTable table,
                                                double fraction,
                                                ExecutionMonitor exec)
                                         throws CanceledExecutionException

Creates a filter that to filter the first 100 * fraction rows from a table. The row counter is determined based on the row number of table.

Parameters:

table - the table from which to get the final row count

fraction - the fraction of the row count that shall survive

exec - an execution monitor to check for cancelation

Returns:

a row filter for this purpose

Throws:

CanceledExecutionException - if exec cancels the row counting

createRangeFilter

public static final RowFilter createRangeFilter(int count)

Creates a filter that passes only the first count rows.

Parameters:

count - the number of rows that survive (starting from top)

Returns:

a filter that only filter the first count rows

createSampleFilter

public static final RowFilter createSampleFilter(double fraction)

Creates row filter that randomly samples about 100 * fraction percent from a table.

Parameters:

fraction - the fraction being used, must be in [0, 1]

Returns:

such a filter

See Also:

RandomFractionRowFilter.RandomFractionRowFilter(double)

createSampleFilter

public static final RowFilter createSampleFilter(DataTable table,
                                                 double fraction,
                                                 ExecutionMonitor exec)
                                          throws CanceledExecutionException

Creates row filter that samples precisely a given fraction of rows. This requires a scan on the table in order to count the rows and determine the right number of sampled rows.

Parameters:

table - to count rows on

fraction - the fraction to be sampled, must be in [0, 1]

exec - to check canceled status on and report progress

Returns:

such a filter

Throws:

CanceledExecutionException - if canceled

createSampleFilter

public static final RowFilter createSampleFilter(DataTable table,
                                                 double fraction,
                                                 Random rand,
                                                 ExecutionMonitor exec)
                                          throws CanceledExecutionException

Creates row filter that samples precisely a given fraction of rows. This requires a scan on the table in order to count the rows and determine the right number of sampled rows. A given Random object makes the sampling "deterministic".

Parameters:

table - to count rows on

fraction - the fraction to be sampled, must be in [0, 1]

rand - the random object for controlled sampling. (If null, uses default)

exec - to check canceled status on and report progress.

Returns:

such a filter

Throws:

CanceledExecutionException - if canceled

createSampleFilter

public static final RowFilter createSampleFilter(DataTable table,
                                                 int count,
                                                 ExecutionMonitor exec)
                                          throws CanceledExecutionException

Creates row filter that samples arbitrary count rows from table.

Parameters:

table - the table from which to create the sample

count - the number of rows the should go "through" the filter

exec - an execution monitor to check for cancelation (this method requires an iteration over table - which might take long)

Returns:

a row ilter to be used for this kind of sampling.

Throws:

CanceledExecutionException - if exec was canceled

createSampleFilter

public static final RowFilter createSampleFilter(DataTable table,
                                                 int count,
                                                 Random rand,
                                                 ExecutionMonitor exec)
                                          throws CanceledExecutionException

Creates row filter that samples arbitrary count rows from table. A given Random object makes the sampling "deterministic".

Parameters:

table - the table from which to create the sample

count - the number of rows the should go "through" the filter

rand - the random object for controlled sampling. (If null, uses default)

exec - an execution monitor to check for cancelation (this method requires an iteration over table - which might take long)

Returns:

a row filter to be used for this kind of sampling

Throws:

CanceledExecutionException - if exec was canceled