S-Curves in Statistical Forecasting

This model is similar to the modified exponential trend model in that it assumes that the underlying trend is gradually approaching a saturation level, specified by the user. The main difference here is the ability of this model to effectively deal with the whole of a typical product life cycle from the launch/rapid growth stage through to saturation level.

Hence, the model may be used at any pre-decline stage of a product's life cycle, assuming that at least 12 months of historical demand data are available. The model uses the classical S-shaped curve to model the underlying trend.

• IFP will obtain a trend curve of overall best fit to the adjusted historical data, taking into account the specified saturation level
• This trend curve is then projected into the future
• All selected historical periods are given equal weight
• Forecast values are then obtained by multiplying the projected trend value by the average seasonal factor for each month

Saturation Level

Saturation level must be entered in terms of average units per month, e.g. a saturation level of 100,000 units per month would be equivalent to an annual saturation level of 1,200,000 units per annum.  Note that the demand for a particular month may exceed the specified saturation level due to seasonal effects.