CrossfoldTask (LensKit 2.2.1)

java.lang.Object
- com.google.common.util.concurrent.AbstractFuture<T>
- - org.grouplens.lenskit.eval.AbstractTask<List<TTDataSet>>
  - - org.grouplens.lenskit.eval.data.crossfold.CrossfoldTask

All Implemented Interfaces:

ListenableFuture<List<TTDataSet>>, Future<List<TTDataSet>>, EvalTask<List<TTDataSet>>
```
public class CrossfoldTask
extends AbstractTask<List<TTDataSet>>
```
The command to build and run a crossfold on the data source file and output the partition files

Constructor Summary

Constructors
Constructor and Description

CrossfoldTask()

CrossfoldTask(String n)

Constructors
Constructor and Description
`CrossfoldTask()`
`CrossfoldTask(String n)`

Method Summary

All Methods Instance Methods Concrete Methods Deprecated Methods
Modifier and Type	Method and Description
`protected void`	`createTTFiles()` Write train-test split files
`protected File[]`	`getFiles(String pattern)` Get the list of files satisfying the specified name pattern
`boolean`	`getForce()`
`Holdout`	`getHoldout()`
`boolean`	`getIsolate()` Query whether this task will produce isolated data sets.
`CrossfoldMethod`	`getMethod()` Get the method to be used for crossfolding.
`String`	`getName()` Get the visible name of this crossfold split.
`int`	`getPartitionCount()` Get the number of folds.
`int`	`getSampleSize()`
`DataSource`	`getSource()` Get the data source backing this crossfold manager.
`String`	`getTestPattern()`
`String`	`getTrainPattern()`
`List<TTDataSet>`	`getTTFiles()` Get the train-test splits as data sets.
`boolean`	`getWriteTimestamps()` Query whether timestamps will be written.
`protected DataSource`	`makeDataSource(File file)`
`protected RatingWriter`	`makeWriter(File file)`
`List<TTDataSet>`	`perform()` Run the crossfold command.
`CrossfoldTask`	`setCache(boolean on)` Configure whether the data sets created by the crossfold will have caching turned on.
`CrossfoldTask`	`setForce(boolean force)` Set the force running option of the command.
`CrossfoldTask`	`setHoldout(int n)` Set holdout to a fixed number of items per user.
`CrossfoldTask`	`setHoldoutFraction(double f)` Set holdout to a fraction of each user's profile.
`CrossfoldTask`	`setIsolate(boolean on)` Configure whether the train-test data sets generated by this task will be isolated.
`CrossfoldTask`	`setMethod(CrossfoldMethod m)` Set the crossfold method.
`CrossfoldTask`	`setOrder(Order<Rating> o)` Set the order for the train-test splitting.
`CrossfoldTask`	`setPartitions(int partition)` Set the number of partitions to generate.
`CrossfoldTask`	`setRetain(int n)` Set holdout from using the retain part to a fixed number of items.
`CrossfoldTask`	`setSampleSize(int n)` Set the sample size (# of users sampled per partition).
`CrossfoldTask`	`setSource(DataSource source)` Set the input data source.
`void`	`setSplitUsers(boolean splitUsers)` Deprecated. Use `setMethod(CrossfoldMethod)` instead.
`CrossfoldTask`	`setTest(String pat)` Set the pattern for the test set files.
`CrossfoldTask`	`setTrain(String pat)` Set the pattern for the training set files.
`CrossfoldTask`	`setWriteTimestamps(boolean pack)` Configure whether to include timestamps in the output file.
`protected Long2IntMap`	`splitUsers(UserDAO dao)` Split users ids to n splits, where n is the partitionCount
`String`	`toString()`
`protected void`	`writeRating(TableWriter writer, Rating rating)` Writing a rating event to the file using table writer
`protected void`	`writeTTFilesByRatings(RatingWriter[] trainWriters, RatingWriter[] testWriters)` Write the split files by Ratings from the DAO
`protected void`	`writeTTFilesByUsers(RatingWriter[] trainWriters, RatingWriter[] testWriters)` Write the split files by Users from the DAO using specified holdout method

Methods inherited from class org.grouplens.lenskit.eval.AbstractTask
execute, getProject, setName, setProject

Methods inherited from class com.google.common.util.concurrent.AbstractFuture
addListener, cancel, get, get, interruptTask, isCancelled, isDone, set, setException, wasInterrupted

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface com.google.common.util.concurrent.ListenableFuture
addListener

Methods inherited from interface java.util.concurrent.Future
cancel, get, get, isCancelled, isDone

- Constructor Detail
  - CrossfoldTask
```
public CrossfoldTask()
```
  - CrossfoldTask
```
public CrossfoldTask(String n)
```
- Method Detail
  - setPartitions
```
public CrossfoldTask setPartitions(int partition)
```
    Set the number of partitions to generate.
    
    Parameters:
    
    partition - The number of paritions
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
  - getSampleSize
```
public int getSampleSize()
```
  - setSampleSize
```
public CrossfoldTask setSampleSize(int n)
```
    Set the sample size (# of users sampled per partition). Only meaningful when the method is CrossfoldMethod.SAMPLE_USERS.
    
    Parameters:
    
    n - The number of users to sample for each partition.
    
    Returns:
    
    The task (for chaining).
  - setTrain
```
public CrossfoldTask setTrain(String pat)
```
    Set the pattern for the training set files. The pattern should have a single format conversion capable of taking an integer ('%s' or '%d') which will be replaced with the fold number.
    
    Parameters:
    
    pat - The training file name pattern.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
    
    See Also:
    
    String.format(String, Object...)
  - setTest
```
public CrossfoldTask setTest(String pat)
```
    Set the pattern for the test set files.
    
    Parameters:
    
    pat - The test file name pattern.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
    
    See Also:
    
    setTrain(String)
  - setOrder
```
public CrossfoldTask setOrder(Order<Rating> o)
```
    Set the order for the train-test splitting. To split a user's ratings, the ratings are first ordered by this order, and then partitioned.
    
    Parameters:
    
    o - The sort order.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
    
    See Also:
    
    RandomOrder, TimestampOrder, setHoldoutFraction(double), setHoldout(int)
  - setHoldout
```
public CrossfoldTask setHoldout(int n)
```
    Set holdout to a fixed number of items per user. Only meaningful when the method is CrossfoldMethod.PARTITION_USERS.
    
    Parameters:
    
    n - The number of items to hold out from each user's profile.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
  - setRetain
```
public CrossfoldTask setRetain(int n)
```
    Set holdout from using the retain part to a fixed number of items. Only meaningful when the method is CrossfoldMethod.PARTITION_USERS.
    
    Parameters:
    
    n - The number of items to train data set from each user's profile.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
  - setHoldoutFraction
```
public CrossfoldTask setHoldoutFraction(double f)
```
    Set holdout to a fraction of each user's profile. Only meaningful when the method is CrossfoldMethod.PARTITION_USERS.
    
    Parameters:
    
    f - The fraction of a user's ratings to hold out.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
  - setSource
```
public CrossfoldTask setSource(DataSource source)
```
    Set the input data source.
    
    Parameters:
    
    source - The data source to use.
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
  - setForce
```
public CrossfoldTask setForce(boolean force)
```
    Set the force running option of the command. The crossfold will be forced to ran with the isForced set to true regardless of whether the partition files are up to date.
    
    Parameters:
    
    force - The force to run option
    
    Returns:
    
    The CrossfoldCommand object (for chaining)
  - setSplitUsers
```
@Deprecated
public void setSplitUsers(boolean splitUsers)
```
    Deprecated. Use setMethod(CrossfoldMethod) instead.
    
    Configure whether it splits per-user or per-rating.
    
    Parameters:
    
    splitUsers - true to split by users (CrossfoldMethod.PARTITION_USERS), false to split by rating (CrossfoldMethod.PARTITION_RATINGS).
  - getMethod
```
public CrossfoldMethod getMethod()
```
    Get the method to be used for crossfolding.
    
    Returns:
    
    The configured crossfold method.
  - setMethod
```
public CrossfoldTask setMethod(CrossfoldMethod m)
```
    Set the crossfold method. The default is CrossfoldMethod.PARTITION_USERS.
    
    Parameters:
    
    m - The crossfold method to use.
  - setCache
```
public CrossfoldTask setCache(boolean on)
```
    Configure whether the data sets created by the crossfold will have caching turned on.
    
    Parameters:
    
    on - Whether the data sets returned should cache.
    
    Returns:
    
    The command (for chaining)
  - setIsolate
```
public CrossfoldTask setIsolate(boolean on)
```
    Configure whether the train-test data sets generated by this task will be isolated. If yes, then each data set will be in its own isolation group; otherwise, they will all be in the default isolation group (the all-zero UUID).
    
    Parameters:
    
    on - true to produce isolated data sets.
    
    Returns:
    
    The task (for chaining).
  - getIsolate
```
public boolean getIsolate()
```
    Query whether this task will produce isolated data sets.
    
    Returns:
    
    true if this task will produce isolated data sets.
  - setWriteTimestamps
```
public CrossfoldTask setWriteTimestamps(boolean pack)
```
    Configure whether to include timestamps in the output file.
    
    Parameters:
    
    pack - true to include timestamps (the default), false otherwise.
    
    Returns:
    
    The task (for chaining).
  - getWriteTimestamps
```
public boolean getWriteTimestamps()
```
    Query whether timestamps will be written.
    
    Returns:
    
    true if output will include timestamps.
  - getName
```
public String getName()
```
    Get the visible name of this crossfold split.
    
    Overrides:
    
    getName in class AbstractTask<List<TTDataSet>>
    
    Returns:
    
    The name of the crossfold split.
  - getTrainPattern
```
public String getTrainPattern()
```
  - getTestPattern
```
public String getTestPattern()
```
  - getSource
```
public DataSource getSource()
```
    Get the data source backing this crossfold manager.
    
    Returns:
    
    The underlying data source.
  - getPartitionCount
```
public int getPartitionCount()
```
    Get the number of folds.
    
    Returns:
    
    The number of folds in this crossfold.
  - getHoldout
```
public Holdout getHoldout()
```
  - getForce
```
public boolean getForce()
```
  - perform
```
public List<TTDataSet> perform()
                        throws TaskExecutionException
```
    Run the crossfold command. Write the partition files to the disk by reading in the source file.
    
    Specified by:
    
    perform in class AbstractTask<List<TTDataSet>>
    
    Returns:
    
    The partition files stored as a list of TTDataSet
    
    Throws:
    
    TaskExecutionException
  - getFiles
```
protected File[] getFiles(String pattern)
```
    Get the list of files satisfying the specified name pattern
    
    Parameters:
    
    pattern - The file name pattern
    
    Returns:
    
    The list of files
  - createTTFiles
```
protected void createTTFiles()
                      throws IOException
```
    Write train-test split files
    
    Throws:
    
    IOException - if there is an error writing the files.
  - writeTTFilesByUsers
```
protected void writeTTFilesByUsers(RatingWriter[] trainWriters,
                                   RatingWriter[] testWriters)
                            throws TaskExecutionException
```
    Write the split files by Users from the DAO using specified holdout method
    
    Parameters:
    
    trainWriters - The tableWriter that write train files
    
    testWriters - The tableWriter that writ test files
    
    Throws:
    
    TaskExecutionException
  - writeTTFilesByRatings
```
protected void writeTTFilesByRatings(RatingWriter[] trainWriters,
                                     RatingWriter[] testWriters)
                              throws TaskExecutionException
```
    Write the split files by Ratings from the DAO
    
    Parameters:
    
    trainWriters - The tableWriter that write train files
    
    testWriters - The tableWriter that writ test files
    
    Throws:
    
    TaskExecutionException
  - writeRating
```
protected void writeRating(TableWriter writer,
                           Rating rating)
                    throws IOException
```
    Writing a rating event to the file using table writer
    
    Parameters:
    
    writer - The table writer to output the rating
    
    rating - The rating event to output
    
    Throws:
    
    IOException - The writer IO error
  - splitUsers
```
protected Long2IntMap splitUsers(UserDAO dao)
```
    Split users ids to n splits, where n is the partitionCount
    
    Parameters:
    
    dao - The DAO of the source file
    
    Returns:
    
    a map of users to partition numbers. Users not in a partition will return -1.
  - getTTFiles
```
public List<TTDataSet> getTTFiles()
```
    Get the train-test splits as data sets.
    
    Returns:
    
    The partition files stored as a list of TTDataSet
  - makeWriter
```
protected RatingWriter makeWriter(File file)
                           throws IOException
```
    Throws:
    
    IOException
  - makeDataSource
```
protected DataSource makeDataSource(File file)
```
  - toString
```
public String toString()
```
    Overrides:
    
    toString in class Object

Class CrossfoldTask

Constructor Summary

Method Summary

Methods inherited from class org.grouplens.lenskit.eval.AbstractTask

Methods inherited from class com.google.common.util.concurrent.AbstractFuture

Methods inherited from class java.lang.Object

Methods inherited from interface com.google.common.util.concurrent.ListenableFuture

Methods inherited from interface java.util.concurrent.Future

Constructor Detail

CrossfoldTask

CrossfoldTask

Method Detail

setPartitions

getSampleSize

setSampleSize

setTrain

setTest

setOrder

setHoldout

setRetain

setHoldoutFraction

setSource

setForce

setSplitUsers

getMethod

setMethod

setCache

setIsolate

getIsolate

setWriteTimestamps

getWriteTimestamps

getName

getTrainPattern

getTestPattern

getSource

getPartitionCount

getHoldout

getForce

perform

getFiles

createTTFiles

writeTTFilesByUsers

writeTTFilesByRatings

writeRating

splitUsers

getTTFiles

makeWriter

makeDataSource

toString