There are two types of forecasting: (1) point forecast, and (2) prediction interval forecast. Point Forecast: A point forecast is a single number that represents our best prediction of the actual value of the variable being forecasted.
Prediction Interval Forecast: A prediction interval forecast
is an interval or range of numbers that is calculated so that we are very confident that the actual value will be contained in the interval.
Forecasting Methods There are many forecasting methods that can be used to predict future events. These methods can be divided into two basic types: Qualitative methods and Quantitative methods.
Qualitative Forecasting Methods
Qualitative forecasting techniques are used when historical data are limited or not available at all. These methods are usually based on the opinions of experts to predict future events subjectively. These methods are required when new product is being introduced. To forecast sales for the new product, a company must rely on expert opinion, which can be supplied by members of its sales force and market research team. There are some commonly used qualitative forecasting techniques. They are also called Judgmental forecasting methods: 1)Subjective curve fitting: In predicting sales of the new product, it is convenient to consider “product life cycle.” This life cycle consist of three stages: a) growth (requires S-curve), b) maturity, and c) decline. The subjective construction of such curve is difficult and requires a great deal of expertise and judgment. 2)Delphi method: The Delphi method assumes that the panel members are recognized experts in the field of interest, and assumes combined knowledge of panel members may produce predictions. Each participant is asked to respond a questionnaire and return it to panel coordinator. 3)Time-independent Technological comparison: In this method, which is often used in predicting technological change, changes in one area are predicted by monitoring changes that take place in another area. The forecaster tries to determine a pattern of change in one area, often called primary trend, which he believes will result in new developments in some other area. Quantitative Forecasting Method Quantitative forecasting methods are used when historical data are available. Quantitative methods can be grouped in two kinds: Univariate models and Casual models. Univariate models: These models predict future values of the variable of interest solely on the basis of the historical pattern (time series) of that variable, assuming that the historical pattern will continue. Univariate models are most useful when conditions are expected to remain the same; they are not very useful in forecasting the impact of changes in management policies, e.g. increase in price or advertising expenditures. Univariate models include – Time series regression, classical decomposition, exponential smoothing, and the Box-Jenkins methodology. Casual models: These models predict future values of the variable of interest based on relationship between that variable to be forecasted and the other variables. That is, the relationship between dependent variable to be forecasted and the other independent variables. The use of casual models involves the identification of other variables that are related to the variable to be predicted. In business world, casual models are advantageous because they allow management to evaluate the impact of various alternative policies. For example, management might wish to predict how various price structures and levels of advertising expenditures will affect sales. On the other hand, casual models are difficult to develop. Moreover, the ability to predict the dependent variable depends on the ability of the forecaster to accurately predict future values independent variables. Choosing a Forecasting Technique Forecasts are generated for points in time that may be a number of days, weeks, months, quarters, or years in the future. The forecaster must consider the following factors when choosing the forecasting technique: 1. The time frame, i.e. immediate, short term, medium, or long term 2. The pattern of data, i.e. the four components of time series (T, C, S, I) 3. The cost of forecasting, i.e. cost of developing the model must be considered 4. The accuracy desired, i.e. some situation forecast limit as much as 20% error as acceptable and in some situations forecast error cannot be more than 1%. 5. The availability of data, i.e. the quantity of available data 6. The ease of operation and understanding, i.e. if the manager does not have confidence in the technique used in predictions, then the predicted values will not be used in the decision making process.