Introduction
Practitioners implement many various forecasting methods, each with its own strengths and weaknesses. These methods include quantitative and qualitative, with the former divided into two segments: one-dimensional and multidimensional. The advantages and disadvantages matter as they need to be reviewed prior to applying a “tool bag” approach to forecasting. By better understanding these methods, choosing the most suitable ones will be easier.
Judgmental Methods
Judgmental methods are used in situations when no other option is available. The advantages of these methods include low development costs, rapid creation, and the ability of executives to properly understand the situation. However, the disadvantages include bias towards the user group, inconsistency over time, and executives’ poor comprehension of the situation.
One-Dimensional Methods
As mentioned earlier, the quantitative methods are divided into one-dimensional and multidimensional segments. One-dimensional ones are formed based on future sales, mimicking the pattern of past sales in the future. They rely on pattern identification within the forecast of past sales history and the assumption that they will continue. The advantages of time series include matching the cases of sales forecasts required for multiple products, products with stable sales, and ease of understanding and use (Chase, 1997). The disadvantages feature a great amount of historical data, poor adjustment to alterations in sales, and they can collapse in long forecast horizons.
Multidimensional Methods
Multidimensional models rely on the premise that upcoming sales of a certain product are related to changes in other variables. Once the nature of the association is quantified, it can be implemented for sales forecasting. The advantages include the availability of software packages, cheap use of computers, and forecast accuracy. However, the disadvantages are dependence on a relationship between independent and dependent variables and poor understanding among managers.
Conclusion
Judgmental methods can be implemented if the products have unstable data. Meanwhile, if the data is stable, though the historical data is incomplete, time series methods are required. If the data is complete and stable, some of the causal methods can be implemented. However, it is important to mention that most of the products in most product portfolios can be predicted by advanced one-dimensional and multidimensional methods.
Reference
Chase Jr., C. W. (1997). Selecting the appropriate forecasting method. Journal of Business Forecasting Methods & Systems, 16(3), 2.