Question

“A recent report produced by Shop n Pay’s chief demand planner has revealed that the retailer’s reliance on simple exponential smoothing to forecast demand for its products is partly responsible of the significant pile-up of goods in the company’s main distribution centre.”

In light of the statement above, highlight any THREE (3) weakness of the simple exponential smoothing method of forecasting.                                                                                                                   

Answer 1

Simple (single) exponential is a time series forecasting method for univariate data that supports data without a trend or seasonal component. It requires a single parameter called alpha, which is the smoothing factor or a smoothing coefficient. This parameter controls the rate at which the influence of prior information will change exponentially. Alpha is set between 0 to 1. The larger alpha value indicates that the model pays more attention to recent past observations but smaller values indicate history is taken into account while making predictions. A value close to 1 indicates fast learning whereas a value close to 0 indicates slow learning.

The main weakness of the simple exponential smoothing are as follows:

1. Exponential smoothing will lag: It means the forecast is behind as the trend will vary i.e. increase or decrease over time. It cannot account for the random variation so the graph will show a smooth line or curve.

2. Exponential smoothing is incapable to account for the dynamic changes at work in the real world. The forecast needs to be constantly updated to account for new information.

3. Exponential smoothing cannot handle trends: Exponential smoothing is best for short term forecasting. It does not take into account seasonal or cyclical variations. Therefore, this forecast is not suitable if there is a trend in the series.

To conclude Exponential smoothing is suitable for short term forecasting as it assumes future trends and patterns will be the same as the current patterns and trends. It assumes a reasonable amount of continuity between the past and future. It ignores random variation in the data.