Chapter 7 Demand Forecasting in a Supply Chain Forecasting -5 Adaptive Trend and Seasonality Adjusted Exponential Smoothing Ardavan Asef-Vaziri References: Supply Chain Management; Chopra and Meindl USC Marshall School of Business Lecture Notes
Ardavan Asef-Vaziri Monthly US Electric Power Consumption Trend and Seasonality: Adaptive -2
Ardavan Asef-Vaziri Trend and Seasonality Trend and Seasonality: Adaptive -3
Ardavan Asef-Vaziri Trend & Seasonality-Corrected Exponential Smoothing Trend and Seasonality: Adaptive -4 The estimates of level, trend, and seasonality are adjusted after each demand observation. Assume periodicity p F t+1 = ( L t + T t )S t+1 = forecast for period t+1 in period t F t+l = ( L t + lT t )S t+l = forecast for period t+l in period t L t = Estimate of level at the end of period t T t = Estimate of trend at the end of period t S t = Estimate of seasonal factor for period t F t = Forecast of demand for period t (made at period t-1 or earlier) D t = Actual demand observed in period t
Ardavan Asef-Vaziri General Steps in Adaptive Forecasting 0- Initialize: Compute initial estimates of level, L 0, trend,T 0, and seasonal factors, S 1,…,S p. As in static forecasting. 1- Forecast: Forecast demand for period t+1 using the general equation, F t+1 = (L t +T t )×S t+1 2- Estimate error: Compute error E t+1 = F t+1 - D t+1 3- Modify estimates: Modify the estimates of level, L t+1, trend, T t+1, and seasonal factor, S t+p+1, given the error E t+1 in the forecast Repeat steps 1, 2, and 3 for each subsequent period Trend and Seasonality: Adaptive -5
Ardavan Asef-Vaziri After observing demand for period t+1, revise estimates for level, trend, and seasonal factors as follows: L t+1 = (D t+1 /S t+1 ) + (1- )(L t +T t ) T t+1 = (L t+1 - L t ) + (1- )T t S t+p+1 = (D t+1 /L t+1 ) + (1- )S t+1 = smoothing constant for level = smoothing constant for trend = smoothing constant for seasonal factor Trend & Seasonality-Corrected Exponential Smoothing
Ardavan Asef-Vaziri Trend & Seasonality-Corrected Exponential Smoothing Example: Tahoe Salt data. Forecast demand for period 1 using Winter’s model. Initial estimates of level, trend, and seasonal factors are obtained as in the static forecasting case L0 = T0 = 524S1=0.47, S2=0.68, S3=1.17, S4=1.66 F1 = (L0 + T0)S1 = ( )(0.47) = 18963(0.47)= 8913 The observed demand for period 1 = D1 = Assume = 0.1, =0.2, =0.1
Ardavan Asef-Vaziri L1 = (Actual Surrogate) + (1- )(Forecast Surrogate) Forecast Surrogate for L1 = L0+T0 Actual Surrogate for L1 = D1/S1 L1 = (D1/S1) + (1- )(L0+T0) L1 = (D1/S1) + 0.9(L0+T0) L1 =(0.1)(8000/0.47)+(0.9)( )=18769 T1 = (Actual Surrogate) + (1- )(Forecast Surrogate) Forecast Surrogate for T1 = T0 Actual Surrogate for T1 = D1-D0 T1 = (L2-L1) + 0.8(T0) T1 = (0.2)( )+(0.8)(524) = 485 Trend & Seasonality-Corrected Exponential Smoothing
Ardavan Asef-Vaziri S5 = (Actual Surrogate) + (1- )(Forecast Surrogate) Forecast Surrogate for S5 = S1 Actual Surrogate for S5 = D1/L1 S5 = (D1/L1) + (1- )(S1) S5 = (D1/L1) + 0.9(S1) S5 = (0.1)(8000/18769)+(0.9)(0.47) = 0.47 F2 = (L1+T1)S2 = ( )(0.68) = Trend & Seasonality-Corrected Exponential Smoothing
Ardavan Asef-Vaziri L1 = 18769, T1 = 485, S2 = 0.68, D2 = L2 = (D2/S2) + 0.9(L1+T1) D2/S2 = 13000/0.68 = L1+T1 = = L2 = (19118) + 0.9(19254) = T2 = (L2-L1) + 0.8(T1) T1 = (0.2)( )+(0.8)(485) = 482 S5 = (Actual Surrogate) + (1- )(Forecast Surrogate) S6 = (D2/L2) + 0.9(S2) S5 = (0.1)(13000/19240)+(0.9)(0.68) = 0.68 F3 = (L2+T2)S3 = ( )(0.68) = Trend & Seasonality-Corrected Exponential Smoothing
Ardavan Asef-Vaziri Forecasting in Practice Collaborate in building forecasts The value of data depends on where you are in the supply chain Be sure to distinguish between demand and sales
Ardavan Asef-Vaziri Practice: Given L0 = 11, T0 = 1, S1 to S4 =0.5,1.0,1.5,1.0 Trend and Seasonality: Adaptive -12 QuarterDemandForecastLevelTrendSeasonal Forecast 1 = (11+1)*0.5
Ardavan Asef-Vaziri L1, T1, F2, S5 Trend and Seasonality: Adaptive -13 QuarterDemandForecastLevelTrendSeasonal New level = 0.25(6/0.5)+0.75(11+1)=12 New trend = 0.25(12-11)+0.75(1)=1 New seasonal = 0.25(6/12)+0.75(0.5)=0.5 New Forecast = (12+1)*1=13
Ardavan Asef-Vaziri L2, T2, F3, S6 Trend and Seasonality: Adaptive -14 QuarterDemandForecastLevelTrendSeasonal New level = 0.25(14/1)+0.75*(12+1)=13.25 New trend = 0.25( )+0.75(1)=1.06 New seasonal = 0.25(14/13.25)+0.75*1=1.014 New Forecast = ( )*1.5=21.45
Ardavan Asef-Vaziri Practice: α = 0.05, β = 0.1, δ = 0.1
Ardavan Asef-Vaziri Assignment Trend and Seasonality: Adaptive -16 Each cycle is 4 periods long. Periodicity = 4. There are three cycles. Compute b0, b1, S1, S2, S3, S4 using static method and forecast using trend and seasonality adjusted method for α= β = δ = 0.25
Ardavan Asef-Vaziri Using Static Model We Can Compute Seasonality Trend and Seasonality: Adaptive -17 b0 (Level) and b1 (Trend) are computed exactly the same as static method by applying regression on deseasonalized data. Initial average seasonality indices are also computed in the same way.
Ardavan Asef-Vaziri Practice; α=β= γ = 0.25 Trend and Seasonality: Adaptive -18
Ardavan Asef-Vaziri Practice; α=β= γ = 0.25 Trend and Seasonality: Adaptive -19