org.elasticsearch.search.aggregations.pipeline.movavg.models

Class HoltWintersModel

java.lang.Object

org.elasticsearch.search.aggregations.pipeline.movavg.models.MovAvgModel

org.elasticsearch.search.aggregations.pipeline.movavg.models.HoltWintersModel

public class HoltWintersModel
extends MovAvgModel

Calculate a triple exponential weighted moving average

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	HoltWintersModel.HoltWintersModelBuilder
static class	HoltWintersModel.HoltWintersModelParser
static class	HoltWintersModel.SeasonalityType

Nested classes/interfaces inherited from class org.elasticsearch.search.aggregations.pipeline.movavg.models.MovAvgModel

MovAvgModel.AbstractModelParser

Field Summary

Fields

Modifier and Type	Field and Description
protected static ParseField	NAME_FIELD
static MovAvgModelStreams.Stream	STREAM

Constructor Summary

Constructors

Constructor and Description
HoltWintersModel(double alpha, double beta, double gamma, int period, HoltWintersModel.SeasonalityType seasonalityType, boolean pad)

Method Summary

Methods

Modifier and Type	Method and Description
boolean	canBeMinimized() Returns if the model can be cost minimized.
MovAvgModel	clone() Clone the model, returning an exact copy
protected <T extends Number> double[]	doPredict(Collection<T> values, int numPredictions) Predicts the next `n` values in the series, using the smoothing model to generate new values.
boolean	hasValue(int valuesAvailable) Checks to see this model can produce a new value, without actually running the algo.
boolean	minimizeByDefault() Should this model be fit to the data via a cost minimizing algorithm by default?
MovAvgModel	neighboringModel() Generates a "neighboring" model, where one of the tunable parameters has been randomly mutated within the allowed range.
<T extends Number> double	next(Collection<T> values) Returns the next value in the series, according to the underlying smoothing model
<T extends Number> double[]	next(Collection<T> values, int numForecasts) Calculate a doubly exponential weighted moving average
void	writeTo(StreamOutput out) Write the model to the output stream

Methods inherited from class org.elasticsearch.search.aggregations.pipeline.movavg.models.MovAvgModel

emptyPredictions, predict

Methods inherited from class java.lang.Object

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

Field Detail

NAME_FIELD

protected static final ParseField NAME_FIELD

STREAM

public static final MovAvgModelStreams.Stream STREAM

Constructor Detail

HoltWintersModel

public HoltWintersModel(double alpha,
                double beta,
                double gamma,
                int period,
                HoltWintersModel.SeasonalityType seasonalityType,
                boolean pad)

Method Detail

minimizeByDefault

public boolean minimizeByDefault()

Description copied from class: MovAvgModel

Should this model be fit to the data via a cost minimizing algorithm by default?

Overrides:

minimizeByDefault in class MovAvgModel

canBeMinimized

public boolean canBeMinimized()

Description copied from class: MovAvgModel

Returns if the model can be cost minimized. Not all models have parameters which can be tuned / optimized.

Specified by:

canBeMinimized in class MovAvgModel

neighboringModel

public MovAvgModel neighboringModel()

Description copied from class: MovAvgModel

Generates a "neighboring" model, where one of the tunable parameters has been randomly mutated within the allowed range. Used for minimization

Specified by:

neighboringModel in class MovAvgModel

clone

public MovAvgModel clone()

Description copied from class: MovAvgModel

Clone the model, returning an exact copy

Specified by:

clone in class MovAvgModel

hasValue

public boolean hasValue(int valuesAvailable)

Description copied from class: MovAvgModel

Checks to see this model can produce a new value, without actually running the algo. This can be used for models that have certain preconditions that need to be met in order to short-circuit execution

Overrides:

hasValue in class MovAvgModel

Parameters:

valuesAvailable - Number of values in the current window of values

Returns:

Returns `true` if calling next() will produce a value, `false` otherwise

doPredict

protected <T extends Number> double[] doPredict(Collection<T> values,
                                    int numPredictions)

Predicts the next `n` values in the series, using the smoothing model to generate new values. Unlike the other moving averages, HoltWinters has forecasting/prediction built into the algorithm. Prediction is more than simply adding the next prediction to the window and repeating. HoltWinters will extrapolate into the future by applying the trend and seasonal information to the smoothed data.

Specified by:

doPredict in class MovAvgModel

Type Parameters:

T - Type of numeric

Parameters:

values - Collection of numerics to movingAvg, usually windowed

numPredictions - Number of newly generated predictions to return

Returns:

Returns an array of doubles, since most smoothing methods operate on floating points

next

public <T extends Number> double next(Collection<T> values)

Description copied from class: MovAvgModel

Returns the next value in the series, according to the underlying smoothing model

Specified by:

next in class MovAvgModel

Type Parameters:

T - Type of numeric

Parameters:

values - Collection of numerics to movingAvg, usually windowed

Returns:

Returns a double, since most smoothing methods operate on floating points

next

public <T extends Number> double[] next(Collection<T> values,
                               int numForecasts)

Calculate a doubly exponential weighted moving average

Type Parameters:

T - Type T extending Number

Parameters:

values - Collection of values to calculate avg for

numForecasts - number of forecasts into the future to return

Returns:

Returns a Double containing the moving avg for the window

writeTo

public void writeTo(StreamOutput out)
             throws IOException

Description copied from class: MovAvgModel

Write the model to the output stream

Specified by:

writeTo in class MovAvgModel

Parameters:

out - Output stream

Throws:

IOException