org.datavec.api.transform.sequence.window

Class OverlappingTimeWindowFunction

java.lang.Object

org.datavec.api.transform.sequence.window.OverlappingTimeWindowFunction

All Implemented Interfaces:

java.io.Serializable, WindowFunction

public class OverlappingTimeWindowFunction
extends java.lang.Object
implements WindowFunction

A windowing function based on time, with potentially overlapping windows. Time for each entry in the sequence is provided by a Time column
The overlapping nature of the windowing function allows for things such as a window size of 1 day, produced every hour.
Two parameters (and one optional parameter) are necessary to specify how windowing is conducted: - The size of the window (for example, 1 day)
- The separation between window periods (for example, 1 hour)
- The offet for the start/end time of the windows

Example with a window size of 12 hours, with the window separation of 1 hour, we end up with windows as follows:
(0:00 to 12:00), (1:00 to 13:00), (2:00 to 14:00) and so on.
If the offset was set to 15 minutes, windows would instead be at (0:15 to 12:15), (1:15 to 13:15), (2:15 to 14:15) and so on.

Note that the windows generated by this window function need not contain any data - i.e., it can generate empty an empty window if no data occurs in the specified time period.

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	OverlappingTimeWindowFunction.Builder

Constructor Summary

Constructors

Constructor and Description
OverlappingTimeWindowFunction(java.lang.String timeColumn, long windowSize, java.util.concurrent.TimeUnit windowSizeUnit, long windowSeparation, java.util.concurrent.TimeUnit windowSeparationUnit) Constructor with zero offset
OverlappingTimeWindowFunction(java.lang.String timeColumn, long windowSize, java.util.concurrent.TimeUnit windowSizeUnit, long windowSeparation, java.util.concurrent.TimeUnit windowSeparationUnit, boolean addWindowStartTimeColumn, boolean addWindowEndTimeColumn) Constructor with zero offset, ability to add window start/end time columns
OverlappingTimeWindowFunction(java.lang.String timeColumn, long windowSize, java.util.concurrent.TimeUnit windowSizeUnit, long windowSeparation, java.util.concurrent.TimeUnit windowSeparationUnit, long offset, java.util.concurrent.TimeUnit offsetUnit) Constructor with optional offset
OverlappingTimeWindowFunction(java.lang.String timeColumn, long windowSize, java.util.concurrent.TimeUnit windowSizeUnit, long windowSeparation, java.util.concurrent.TimeUnit windowSeparationUnit, long offset, java.util.concurrent.TimeUnit offsetUnit, boolean addWindowStartTimeColumn, boolean addWindowEndTimeColumn, boolean excludeEmptyWindows) Constructor with optional offset, ability to add window start/end time columns

Method Summary

Modifier and Type	Method and Description
java.util.List<java.util.List<java.util.List<Writable>>>	applyToSequence(java.util.List<java.util.List<Writable>> sequence) Apply the windowing function to the given sequence
Schema	getInputSchema()
void	setInputSchema(Schema schema)
java.lang.String	toString()
Schema	transform(Schema inputSchema) Get the output schema, given the input schema.

Methods inherited from class java.lang.Object

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

Constructor Detail

OverlappingTimeWindowFunction

public OverlappingTimeWindowFunction(java.lang.String timeColumn,
                                     long windowSize,
                                     java.util.concurrent.TimeUnit windowSizeUnit,
                                     long windowSeparation,
                                     java.util.concurrent.TimeUnit windowSeparationUnit)

Constructor with zero offset

Parameters:

timeColumn - Name of the column that contains the time values (must be a time column)

windowSize - Numerical quantity for the size of the time window (used in conjunction with windowSizeUnit)

windowSizeUnit - Unit of the time window

windowSeparation - The separation between consecutive window start times (used in conjunction with WindowSeparationUnit)

windowSeparationUnit - Unit for the separation between windows

OverlappingTimeWindowFunction

public OverlappingTimeWindowFunction(java.lang.String timeColumn,
                                     long windowSize,
                                     java.util.concurrent.TimeUnit windowSizeUnit,
                                     long windowSeparation,
                                     java.util.concurrent.TimeUnit windowSeparationUnit,
                                     boolean addWindowStartTimeColumn,
                                     boolean addWindowEndTimeColumn)

Constructor with zero offset, ability to add window start/end time columns

Parameters:

timeColumn - Name of the column that contains the time values (must be a time column)

windowSize - Numerical quantity for the size of the time window (used in conjunction with windowSizeUnit)

windowSizeUnit - Unit of the time window

windowSeparation - The separation between consecutive window start times (used in conjunction with WindowSeparationUnit)

windowSeparationUnit - Unit for the separation between windows

OverlappingTimeWindowFunction

public OverlappingTimeWindowFunction(java.lang.String timeColumn,
                                     long windowSize,
                                     java.util.concurrent.TimeUnit windowSizeUnit,
                                     long windowSeparation,
                                     java.util.concurrent.TimeUnit windowSeparationUnit,
                                     long offset,
                                     java.util.concurrent.TimeUnit offsetUnit)

Constructor with optional offset

Parameters:

timeColumn - Name of the column that contains the time values (must be a time column)

windowSize - Numerical quantity for the size of the time window (used in conjunction with windowSizeUnit)

windowSizeUnit - Unit of the time window

windowSeparation - The separation between consecutive window start times (used in conjunction with WindowSeparationUnit)

windowSeparationUnit - Unit for the separation between windows

offset - Optional offset amount, to shift start/end of the time window forward or back

offsetUnit - Optional offset unit for the offset amount.

OverlappingTimeWindowFunction

public OverlappingTimeWindowFunction(java.lang.String timeColumn,
                                     long windowSize,
                                     java.util.concurrent.TimeUnit windowSizeUnit,
                                     long windowSeparation,
                                     java.util.concurrent.TimeUnit windowSeparationUnit,
                                     long offset,
                                     java.util.concurrent.TimeUnit offsetUnit,
                                     boolean addWindowStartTimeColumn,
                                     boolean addWindowEndTimeColumn,
                                     boolean excludeEmptyWindows)

Constructor with optional offset, ability to add window start/end time columns

Parameters:

timeColumn - Name of the column that contains the time values (must be a time column)

windowSize - Numerical quantity for the size of the time window (used in conjunction with windowSizeUnit)

windowSizeUnit - Unit of the time window

windowSeparation - The separation between consecutive window start times (used in conjunction with WindowSeparationUnit)

windowSeparationUnit - Unit for the separation between windows

offset - Optional offset amount, to shift start/end of the time window forward or back

offsetUnit - Optional offset unit for the offset amount.

addWindowStartTimeColumn - If true: add a time column (name: "windowStartTime") that contains the start time of the window

addWindowEndTimeColumn - If true: add a time column (name: "windowEndTime") that contains the end time of the window

excludeEmptyWindows - If true: exclude any windows that don't have any values in them

Method Detail

setInputSchema

public void setInputSchema(Schema schema)

Specified by:

setInputSchema in interface WindowFunction

getInputSchema

public Schema getInputSchema()

Specified by:

getInputSchema in interface WindowFunction

Returns:

transform

public Schema transform(Schema inputSchema)

Description copied from interface: WindowFunction

Get the output schema, given the input schema. Typically the output schema is the same as the input schema, but not necessarily (for example, if the window function adds columns for the window start/end times)

Specified by:

transform in interface WindowFunction

Parameters:

inputSchema - Schema of the input data

Returns:

Schema of the output windows

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

applyToSequence

public java.util.List<java.util.List<java.util.List<Writable>>> applyToSequence(java.util.List<java.util.List<Writable>> sequence)

Description copied from interface: WindowFunction

Apply the windowing function to the given sequence

Specified by:

applyToSequence in interface WindowFunction

Parameters:

sequence - the input sequence

Returns:

the sequence with the window function applied