Table 16.6 Minitab Calculations for Consumer Credit Outstanding (a = 0.7.B = 0.6) TIME OBSERVED CONSUMER CREDIT LEVEL EXPECTED VALUE TREND FORECASTS 1 28 27 2 3 19 4 5 6 133 155 165 171 194 231 274 312 313 333 343 130 156 170 177 192 224 266 309 324 338 347 12 14 24 35 40 25 7 8 9 10 18 13 11 12 13 v,8 360 373 385 14 Forecasting Seasonal Time Series We now examine an extension of the Holt-Winters method that allows for seasonality. In most practical problems, the seasonal factor is taken to be multiplicative, so that, for example, in dealing with monthly sales figures, we might think of January in terms of a proportion of average monthly sales. As before, the trend component is assumed to be additive. Similar to the nonseasonal case, we use xp, it, and T, to denote, respectively, the ob- served value and the level and trend estimates at time t. The seasonal factor is denoted F, so if the time series contains s periods per year, the seasonal factor for the corresponding period in the previous year will be F:--- In the Holt-Winters model the estimates of level, trend, and the seasonal factor are updated by the following three equations , = (1 - a)(x-1 + T-1) + a (0