The exponential smoothing model for forecasting does not eliminate any past data, but adjusts the weights given to past data that the older data gets increasingly less weight hence the name exponential smoothing.
The basic idea is each forecast is based on an average which is corrected each time there is a forecast error.
For example, if we forecast 90 units of demand for an item in a particular period and the actual demand for that item is 100 units, we would perhaps increase our forecast by that margin of error which in this case is 10 units. In this way if the error indicated that demand was changing, we would begin to change the forecast.
We may not want to incorporate the entire error since the error may have just been due to the random variations around the mean.
This forecasting model assumes demand is essentially constant, with only random variations around an average, if we have any indication of a pattern underlying the randomness of demand, it is important that it be taken into account to improve our forecast accuracy.
