Demand forecasting refers to the prediction of what will transpire to the current product sales of a company. It is normally done using multi-functional technique. During this process, the input from marketing and sales, production, and finance need to be regarded. The final forecasting of demand involves the consensus of all managers participating in the forecasting process. The process main stages include choosing the products to be forecast, establishing forecast time horizon, choosing forecasting model, collecting data, making forecast, and finally implementing and validating the results. The testing will determine three types of errors that include scale-dependent errors, percentage errors, and scaled error to establish the model accuracy with a model recording minimal errors being considered as the most efficient in forecasting.
The forecasts accuracy can only be established by evaluating how efficient a model can perform on novel data which were not utilized during the model estimation. When selecting models, it is normal to utilize a part of the accessible data for testing and to utilize the remaining data for the model estimation or training. The testing data can then be utilized to evaluate how effective the model is probable to forecast on novel data (Hyndman, 2014).
Qualitative technique of forecasting generally use the experts’ judgment to create forecasts. The main advantage of these processes is that they can be used in places where historical data are not accessible. The three most essential techniques of qualitative methods include the subject approach, scenario writing and Delphi method.