Serialized Form


Package weka.classifiers.timeseries

Class weka.classifiers.timeseries.AbstractForecaster extends java.lang.Object implements Serializable

serialVersionUID: 5179667114364013750L

Serialized Fields

m_fieldsToForecast

java.util.List<E> m_fieldsToForecast
The name of the attribute (class) to forecast

Class weka.classifiers.timeseries.ErrorBasedConfidenceLimitEstimator extends java.lang.Object implements Serializable

serialVersionUID: -2748314799535071043L

Serialized Fields

m_confidenceLimitsForTargets

java.util.List<E> m_confidenceLimitsForTargets
Holds confidence bound offsets for targets at a certain level. The outer list corresponds to the fields to forecast (in the same order as supplied to the TSForecaster.setFieldsToForecast() method. The inner lists hold two element arrays of doubles (upper, lower bounds). The first index of these lists are bounds for to a one-step-ahead forecast, the second a two-step-ahead forecast, and so on


m_targetFields

java.util.List<E> m_targetFields
The names of the target field(s)


m_confidenceLevel

double m_confidenceLevel
The confidence level for the limits

Class weka.classifiers.timeseries.TSLagMaker extends java.lang.Object implements Serializable

serialVersionUID: -1697901820770907975L

Serialized Fields

m_fieldsToLag

java.util.List<E> m_fieldsToLag
The names of the fields to create lagged variables for


m_minLag

int m_minLag
The minimum lag


m_maxLag

int m_maxLag
The maximum lag


m_lagFineTune

java.lang.String m_lagFineTune
Optionally fine tune the selection of lagged attributes within the min and max via a range string.


m_averageConsecutiveLongLags

boolean m_averageConsecutiveLongLags
Whether to replace a number of consecutive long lagged variables with their average.


m_averageLagsAfter

int m_averageLagsAfter
If replacing long lagged variables with averages, do so for those long lagged variables with lag greater than this


m_numConsecutiveToAverage

int m_numConsecutiveToAverage
How many consecutive lagged variables to average, if averaging long lagged variables


m_timeStampName

java.lang.String m_timeStampName
The name of the timestamp attribute (if there is one)


m_adjustForTrends

boolean m_adjustForTrends
Whether to adjust for trends. If a timestamp attribute is named then adjusting for trends will occur. If there is no timestamp attribute in the data, then turning this on will result in an artificial timestamp attribute getting added to the data.


m_adjustForVariance

boolean m_adjustForVariance
Whether to stabilize the variance in the field to be forecast by applying a log transform


m_useArtificialTimeIndex

boolean m_useArtificialTimeIndex
True if an artificial time index has been added to the data


m_includeTimeLagCrossProducts

boolean m_includeTimeLagCrossProducts
Include time/lag interaction terms?


m_lastTimeValue

double m_lastTimeValue
artificial time and last known real time value


m_artificialTimeMaker

AddID m_artificialTimeMaker
Used to add an artificial time attribute to the data if the user has selected to adjust for trends and there isn't a time stamp in the data


m_varianceAdjusters

java.util.List<E> m_varianceAdjusters
Filters for creating the various lagged and derived attributes


m_lagMakers

java.util.List<E> m_lagMakers

m_averagedLagMakers

java.util.List<E> m_averagedLagMakers

m_timeIndexMakers

java.util.List<E> m_timeIndexMakers

m_timeLagCrossProductMakers

java.util.List<E> m_timeLagCrossProductMakers

m_extraneousAttributeRemover

Remove m_extraneousAttributeRemover

m_primaryPeriodicName

java.lang.String m_primaryPeriodicName
The name of the primary periodic attribute


m_primaryPeriodicSequence

java.util.Map<K,V> m_primaryPeriodicSequence
Holds a map of primary periodic values as keys and their immediate successors (chronologically) as values. The primary periodic attribute (if available) should relate to the time interval of the instances (e.g. hourly, daily, monthly etc.).


m_secondaryPeriodicLookups

java.util.Map<K,V> m_secondaryPeriodicLookups
A map (keyed by attribute) of maps for looking up the values of secondary periodic attribute values that correspond to the values of the primary periodic attribute


m_originalHeader

Instances m_originalHeader

m_lastHistoricInstance

Instance m_lastHistoricInstance
This holds the most recent (time wise) training or primed instance. We can use it to determine the t+1 periodic value for the primary periodic attribute


m_am

boolean m_am
pre-defined fields that can be derived from a genuine date time stamp


m_dayOfWeek

boolean m_dayOfWeek

m_weekend

boolean m_weekend

m_monthOfYear

boolean m_monthOfYear

m_quarter

boolean m_quarter

m_customPeriodics

java.util.Map<K,V> m_customPeriodics
custom defined fields that can be derived from a genuine date time stamp


m_derivedPeriodicMakers

java.util.List<E> m_derivedPeriodicMakers

m_dateBasedPeriodicity

TSLagMaker.Periodicity m_dateBasedPeriodicity

m_deleteMissingFromStartOfSeries

boolean m_deleteMissingFromStartOfSeries
Delete instances from the start of the transformed series where lagged variables are missing? Default leaves missing value handling to the base learner.


m_dateTimeStampBase

long m_dateTimeStampBase
Stores the first time stamp value in the data


m_addDateMap

Add m_addDateMap

m_deltaTime

double m_deltaTime
Holds the difference between the time stamps for the two most recent training instances or the average difference over consecutive training instances if the differences are not constant. Either this or date arithmetic (to advance time stamp by month) is used to advance the timestamp for future instances.

Class weka.classifiers.timeseries.WekaForecaster extends AbstractForecaster implements Serializable

serialVersionUID: 5562710925011828590L

Serialized Fields

m_originalHeader

Instances m_originalHeader
The format of the original incoming instances


m_transformedHeader

Instances m_transformedHeader
The format of the transformed data


m_forecaster

Classifier m_forecaster
The base regression scheme to use


m_singleTargetForecasters

java.util.List<E> m_singleTargetForecasters
The individual forecasters for each target


m_modelBuilt

boolean m_modelBuilt
True if the forecaster has been built


m_useArtificialTimeIndex

boolean m_useArtificialTimeIndex
True if an artificial time index has been added to the data


m_confidenceLimitEstimator

ErrorBasedConfidenceLimitEstimator m_confidenceLimitEstimator
The estimator used for calculating confidence limits.


m_calculateConfLimitsSteps

int m_calculateConfLimitsSteps
Number of steps ahead to calculate confidence limits for (0 = don't calculate confidence limits


m_confidenceLevel

double m_confidenceLevel
Confidence level to compute confidence limits at


m_lagMaker

TSLagMaker m_lagMaker
The lag maker to use


m_dateRemover

RemoveType m_dateRemover
For removing any date attributes (TSLagMaker will remap date timestamps to numeric)


m_missingTargetList

java.util.List<E> m_missingTargetList
Holds a list of training instance indexes that contained missing target values that were replaced via interpolation


m_missingTimeStampList

java.util.List<E> m_missingTimeStampList
Holds a list of training instance indexes that contained missing date values (if a date time stamp is being used)

Class weka.classifiers.timeseries.WekaForecaster.SingleTargetForecaster extends java.lang.Object implements Serializable

serialVersionUID: -4404412501006669036L

Serialized Fields

m_targetForecaster

Classifier m_targetForecaster

m_otherTargetRemover

Remove m_otherTargetRemover

m_classIndex

int m_classIndex

m_className

java.lang.String m_className

Package weka.classifiers.timeseries.core

Class weka.classifiers.timeseries.core.AbstractTimeSeriesFilter extends Filter implements Serializable

serialVersionUID: -3795656792078022357L

Serialized Fields

m_SelectedCols

Range m_SelectedCols
Stores which columns to copy


m_FillWithMissing

boolean m_FillWithMissing
True if missing values should be used rather than removing instances where the translated value is not known (due to border effects).


m_InstanceRange

int m_InstanceRange
The number of instances forward to translate values between. A negative number indicates taking values from a past instance.


m_History

java.util.LinkedList<E> m_History
Stores the historical instances to copy values between

Class weka.classifiers.timeseries.core.CustomPeriodicTest extends java.lang.Object implements Serializable

serialVersionUID: -1470297629040276248L

Serialized Fields

m_lowerBoundTest

java.lang.String m_lowerBoundTest

m_upperBoundTest

java.lang.String m_upperBoundTest

m_upperTest

weka.classifiers.timeseries.core.CustomPeriodicTest.TestPart m_upperTest

m_lowerTest

weka.classifiers.timeseries.core.CustomPeriodicTest.TestPart m_lowerTest

Class weka.classifiers.timeseries.core.CustomPeriodicTest.TestPart extends java.lang.Object implements Serializable

serialVersionUID: -898852333853148631L

Serialized Fields

m_boundOperator

weka.classifiers.timeseries.core.CustomPeriodicTest.Operator m_boundOperator

m_year

int m_year

m_month

int m_month

m_week_of_yr

int m_week_of_yr

m_week_of_month

int m_week_of_month

m_day_of_yr

int m_day_of_yr

m_day_of_month

int m_day_of_month

m_day_of_week

int m_day_of_week

m_hour_of_day

int m_hour_of_day

m_min_of_hour

int m_min_of_hour

m_second

int m_second

c

java.util.Calendar c

Class weka.classifiers.timeseries.core.TimeSeriesTranslate extends AbstractTimeSeriesFilter implements Serializable

serialVersionUID: -4799796255517698151L


Package weka.classifiers.timeseries.gui

Class weka.classifiers.timeseries.gui.AdvancedConfigPanel extends javax.swing.JPanel implements Serializable

Serialized Fields

m_instances

Instances m_instances
The training instances to operate on


m_simpleConfig

SimpleConfigPanel m_simpleConfig
A reference to the simple config panel


m_configHolder

javax.swing.JTabbedPane m_configHolder

m_baseLearnerEditor

GenericObjectEditor m_baseLearnerEditor
Editor for selecting and configuring the base algorithm


m_baseLearnerPanel

PropertyPanel m_baseLearnerPanel

m_useCustomLags

javax.swing.JCheckBox m_useCustomLags

m_minLagSpinner

javax.swing.JSpinner m_minLagSpinner

m_maxLagSpinner

javax.swing.JSpinner m_maxLagSpinner

m_adjustForVarianceCheckBox

javax.swing.JCheckBox m_adjustForVarianceCheckBox

m_fineTuneLagsField

javax.swing.JTextField m_fineTuneLagsField

m_averageLongLags

javax.swing.JCheckBox m_averageLongLags

m_averageLagsAfter

javax.swing.JSpinner m_averageLagsAfter

m_numConsecutiveToAverage

javax.swing.JSpinner m_numConsecutiveToAverage

m_dateDerivedPeriodicsHeader

Instances m_dateDerivedPeriodicsHeader

m_customizeDateDerivedPeriodics

javax.swing.JCheckBox m_customizeDateDerivedPeriodics

m_dateDerivedPeriodicSelector

AttributeSelectionPanel m_dateDerivedPeriodicSelector

m_primaryPeriodicCombo

javax.swing.JComboBox m_primaryPeriodicCombo

m_evaluationModsHeader

Instances m_evaluationModsHeader

m_evaluationMetrics

AttributeSelectionPanel m_evaluationMetrics

m_trainingCheckBox

javax.swing.JCheckBox m_trainingCheckBox

m_holdoutCheckBox

javax.swing.JCheckBox m_holdoutCheckBox

m_holdoutSize

javax.swing.JTextField m_holdoutSize

m_allowSeparateTestSet

boolean m_allowSeparateTestSet
show the separate test set checkbox and button?


m_separateTestSetCheckBox

javax.swing.JCheckBox m_separateTestSetCheckBox

m_testSetBut

javax.swing.JButton m_testSetBut

m_outputPredsCheckBox

javax.swing.JCheckBox m_outputPredsCheckBox

m_outputPredsCombo

javax.swing.JComboBox m_outputPredsCombo

m_outputPredsComboLabel

javax.swing.JLabel m_outputPredsComboLabel

m_outputStepSpinner

javax.swing.JSpinner m_outputStepSpinner

m_outputStepLabel

javax.swing.JLabel m_outputStepLabel

m_outputFutureCheckBox

javax.swing.JCheckBox m_outputFutureCheckBox

m_graphPredsAtStepCheckBox

javax.swing.JCheckBox m_graphPredsAtStepCheckBox

m_graphPredsAtStepSpinner

javax.swing.JSpinner m_graphPredsAtStepSpinner

m_stepLab

javax.swing.JLabel m_stepLab

m_graphTargetForStepsCheckBox

javax.swing.JCheckBox m_graphTargetForStepsCheckBox

m_targetComboLabel

javax.swing.JLabel m_targetComboLabel

m_graphTargetAtStepsCombo

javax.swing.JComboBox m_graphTargetAtStepsCombo

m_stepRange

javax.swing.JTextField m_stepRange

m_graphFutureCheckBox

javax.swing.JCheckBox m_graphFutureCheckBox

Class weka.classifiers.timeseries.gui.ForecastingPanel extends javax.swing.JPanel implements Serializable

Serialized Fields

m_instances

Instances m_instances
The training instances to operate on


m_log

Logger m_log
Panel for logging


m_simpleConfigPanel

SimpleConfigPanel m_simpleConfigPanel
The simple configuration panel


m_advancedConfigPanel

AdvancedConfigPanel m_advancedConfigPanel
The advanced configuration panel


m_forecaster

WekaForecaster m_forecaster
The current forecaster


m_configPane

javax.swing.JTabbedPane m_configPane
Tabbed pane to hold the simple and advanced config panels


m_outText

javax.swing.JTextArea m_outText
The main textual output area


m_history

ResultHistoryPanel m_history
A panel holding and controlling the results for viewing


m_startBut

javax.swing.JButton m_startBut
Button to launch the forecaster


m_stopBut

javax.swing.JButton m_stopBut
Button to stop processing


m_splitP

javax.swing.JSplitPane m_splitP

m_runThread

java.lang.Thread m_runThread

m_fileChooser

javax.swing.JFileChooser m_fileChooser
The file chooser for selecting model files.


m_sortedCheck

boolean m_sortedCheck
For each dataset, perform a check (if a timestamp is specified) just once to see if it is in ascending order


m_outputPane

javax.swing.JTabbedPane m_outputPane
Tabbed pane that holds the main text output plus tabs for any generated graphs


m_framedOutputMap

java.util.HashMap<K,V> m_framedOutputMap

Class weka.classifiers.timeseries.gui.SimpleConfigPanel extends javax.swing.JPanel implements Serializable

Serialized Fields

m_instances

Instances m_instances
The training instances to operate on


m_forecaster

WekaForecaster m_forecaster
The forecaster to configure


m_targetHeader

Instances m_targetHeader
Holds the header of the training instances after all but numeric attributes are removed (for target selection purposes


m_timeStampCombo

javax.swing.JComboBox m_timeStampCombo

m_periodicityCombo

javax.swing.JComboBox m_periodicityCombo

m_targetPanel

AttributeSelectionPanel m_targetPanel

m_computeConfidence

javax.swing.JCheckBox m_computeConfidence

m_horizonSpinner

javax.swing.JSpinner m_horizonSpinner

m_confidenceLevelSpinner

javax.swing.JSpinner m_confidenceLevelSpinner

m_performEvaluation

javax.swing.JCheckBox m_performEvaluation

m_advancedConfig

AdvancedConfigPanel m_advancedConfig

m_parentPanel

ForecastingPanel m_parentPanel

Package weka.classifiers.timeseries.gui.explorer

Class weka.classifiers.timeseries.gui.explorer.ExplorerTSPanel extends javax.swing.JPanel implements Serializable

serialVersionUID: -7313227058091543628L

Serialized Fields

m_log

Logger m_log

m_forecastingPanel

ForecastingPanel m_forecastingPanel