Calculation STATBEANS®

 

STATBEAN Name: TimeSeriesForecast

 

Purpose: Time Series Forecasting methods using values of a time series. This STATBEAN allows Statgraphics to work as time series forecasting software.

DataSource: any. 

Read/Write Properties
Name Type Description Possible Values Default Value
arimaAR int Order of nonseasonal autoregressive term in an ARIMA model. 0-4 0
arimaD int Order of nonseasonal differencing in an ARIMA model. 0-2 0
arimaIncludeConstant boolean Whether to include a constant term when estimating ARIMA models. true,false true
arimaMA int Order of nonseasonal moving average term in an ARIMA model. 0-4 0
arimaSAR int Order of seasonal autoregressive term in an ARIMA model. 0-4 0
arimaSD int Order of seasonal differencing in an ARIMA model. 0-2 0
arimaSMA int Order of seasonal moving average term in an ARIMA model. 0-4 0
backforecasting boolean Whether to use backforecasting when estimating ARIMA models. true,false true
confidenceLevel double Level of confidence for the prediction and confidence limits, as a percentage. 0.0001-99.9999 95.0
modelType String Type of model to use for forecasting. "Random Walk","Mean","Linear Trend",
"Quadratic Trend","Exponential Trend",
"SCurve","MA","EWMA",
"Linear Smoothing","Quadratic Smoothing",
"Holts Smoothing","Winters Smoothing"
"ARIMA"
"Random Walk"
movingAverageSpan int Number of terms in the MA smoother. 2+ 5
numberOfForecasts int The number of time periods to forecast. 1+ 12
optimize boolean Whether to optimize the exponential smoothing parameters. true,false true
samplingInterval double The length of time between consecutive data values. Any double > 0.0 1.0
seasonalLength int The number of time periods comprising a season. 1+ 1
smoothingParameter1 double First parameter for exponential smoothing. 0-1, exclusive. 0.1
smoothingParameter2 double Second parameter for exponential smoothing. 0-1, exclusive. 0.1
smoothingParameter3 double Third parameter for exponential smoothing. 0-1, exclusive. 0.1
startTime String The value of time associated with row 1. Any string resulting in the proper type of value. "1.0"
timeScale String The type of time units. "Year","Quarter","Month","Day",
"Hour","Minute","Second","Other"
"Other"
timeSeriesVariableName String The name of the column with data values to be plotted. Any string. ""
withholdForValidation int The number of time periods to withhold from estimation and use for validation statistics. 0+ 0

Other Public Methods For Time Series Analysis Forecasting
Name Description Arguments Return Value
int getArimaDegreesOfFreedom() Returns the residual degrees of freedom for an ARIMA model. None. Degrees of freedom.
int getArimaIterations() Returns the number of iterations used to fit an ARIMA model. None. Iterations.
double getArimaMeanSquaredError() Returns the residual mean squared error when fitting an ARIMA model. None. MSE.
double getArimaParameterEstimate(int i) Returns the estimate of the i-th parameter in an ARIMA model (in the order AR,MA,SAR,SMA,mean). Index. Estimate.
double getArimaStandardError(int i) Returns the standard error of the estimate of the i-th parameter in an ARIMA model. Index. Standard error.
double getForecastData(int row) Returns the forecasted values. Row index (0 origin). Forecasted value.
double getLowerLimit(int row) Returns the lower confidence limits for the forecasts. Row index (0 origin). Lower limit.
double getMAE() Returns the mean absolute error. None. Calculated statistic.
double getMAPE() Returns the mean absolute percentage error. None. Calculated statistic.
double getME() Returns the mean error. None. Calculated statistic.
double getMPE() Returns the mean percentage error. None. Calculated statistic.
double getMSE() Returns the mean squared error. None. Calculated statistic.
double getRMSE() Returns the root mean squared error. None. Calculated statistic.
int getNumberOfMissingValuesReplaced() Returns the number of missing values replaced with estimates. None. Number replaced.
double getOriginalData(int row) Returns the original data values. Row index (0 origin). Original value.
double getResidual(int row) Returns the one step ahead forecast errors. Row index (0 origin). Residual.
int getStatisticsSampleSize() Returns the number of values used to calcualtye the statistics. None. Number of data values.
double getUpperLimit(int row) Returns the upper confidence limits for the forecasts. Row index (0 origin). Upper limit.
double getValidationMAE() Returns the mean absolute error during the validation period. None. Calculated statistic.
double getValidationMAPE() Returns the mean absolute percentage error during the validation period. None. Calculated statistic.
double getValidationME() Returns the mean error during the validation period. None. Calculated statistic.
double getValidationMPE() Returns the mean percentage error during the validation period. None. Calculated statistic.
double getValidationMSE() Returns the mean squared error during the validation period. None. Calculated statistic.
double getValidationRMSE() Returns the root mean squared error during the validation period. None. Calculated statistic.
int getValidationSampleSize() Returns the number of values used to calcualtye the validation statistics. None. Number of data values.

Time Series Forecasting Software Output Variables
Name Description
Residual One step ahead forecast errors.
Forecast Forecasted data values.


Time Series Forecasting Software Code Sample 
- see TimeSeriesForecastPlot.