Why do we need a 7-day Weather Forecast in the first place?

Forecasting provides additional information for planning the future actions based on available data. Different planning problem requires different type of forecasting methods. According to Principle of Business Forecasting (Ord, Fildes and Kourentzes, 2017), a PIVASE framework helps decision makers determine the reasons and forecasting techniques that fits to their needs. PIVASE is an abbreviation of Purpose, Information, Value, Analysis, System, and Evaluation.

Purpose discusses the main goal of forecasting process and its effect on the planning process. Moreover, decision makers also need to set the forecasting horizon, the depth and elements of forecasting itself, and the forecasting origin relevant to the planning process. In Information part PIVASE, the main focus is the data’s availability for the forecasting. Some types of information are sensitive to time and its usefulness could dissipate quickly. Hence, the value of information depends on the impact it brings to the forecasting.

For instance, population changes which affects sales in years from now may irrelevant to forecasting sales in next three to six months. Evaluating value of the forecast also essential prior to the forecasting itself. In other words, the decision makers should consider whether the investment is worth for the knowledge of future events. Analysis focuses on process and techniques used for capturing insights and characteristics within the data.

This includes developing the best forecasting model and comparing using a hold-out sample (data excluded from one for fitting the model). On the other hand, System stage analyze which forecasting system and software that meet the needs (forecasting purpose) of the organizations. This means companies could generate multiple forecasts at different variables simultaneously, hence it is important to provide suitable system for these forecasts.

After generating forecasts from chosen methods using the available system, it is essential to check their effectiveness and search rooms for improvement. The simplest method is comparing past results with the forecasts made and several methods are available based on the difference between actual values and orecast. Each method has its own advantages and disadvantages; thus, it is crucial to assess these methods beforehand since mistakenly using the wrong one could be misleading.