Example of Winters' Method

Budget planners for a local business office want to predict electricity costs for the next year. They collect data on utility usage during the previous 84 billing periods to predict future usage.

  1. Open the sample data, UtilityUsage.MTW.
  2. Choose Stat > Time Series > Winters’ Method.
  3. In Variable, enter Electric Charges.
  4. In Seasonal length, enter 12.
  5. Select Generate forecasts. In Number of forecasts, enter 12.
  6. Click OK.

Interpret the results

On the Winters' method plot, the fits closely follow the data and the seasonal pattern and trend are stable at the end of the data. The budget planners can conclude that the forecasts for the electricity costs are likely to be accurate for the next year.

Method

Model typeMultiplicative Method
DataElectric Charges
Length84

Smoothing Constants

α (level)0.2
γ (trend)0.2
δ (seasonal)0.2

Accuracy Measures

MAPE3.76
MAD62.87
MSD6448.08

Forecasts

PeriodForecastLowerUpper
851607.561453.521761.60
861568.331411.871724.78
871845.461686.322004.61
881990.051827.962152.15
891964.941799.632130.24
901992.041823.292160.78
912231.432059.032403.83
922235.762059.492412.03
932007.061826.722187.39
941937.181752.602121.76
951795.321606.321984.31
961890.921697.362084.48