Forecasting and forecasting methods
General concepts
Predict. It is to issue a statement about what is likely to happen in the future, based on analysis and judgment considerations.
Purpose. To make a forecast is to obtain knowledge about uncertain events that are important in making present decisions.
Forecast: Science and Art
Science. Statistical based methods.
Art. Judgment and intuition about the methodological framework to be used. It involves knowing the environment, selecting the best technique, the number of historical data to be included, etc.
Forecasting techniques reduce uncertainty about the future, allowing the structuring of plans and actions that are consistent with the organization's objectives and also allow taking appropriate and timely corrective actions when situations outside of the forecast occur.
Forecast vs. To plan
Forecast. Advance estimate of the value of a variable, for example: the demand for a product.
Budget. Anticipated value of the variable that a company is able to specify, for example: the amount of product that the company decides to manufacture based on demand and installed capacity.
Knowledge of forecasting techniques is of little value unless they can be applied effectively in the organization's planning process.
Uses of forecasts
• Marketing
- Market size Market share Price trend New product development
• Production
- Raw material cost Labor cost Raw material availability Labor availability Maintenance requirements Available plant capacity for production
• Finance
- Interest rates Slow payment accounts
• Human Resources
- Number of Workers Staff Turnover Trends of Absenteeism Tardy Trend
• Strategic Planning
- Economic factors Price changes Costs Product line growth
Forecast Features
First. All situations in which a forecast is required deal with the future and time is directly involved. Thus, it must be forecast for a specific point in time and changing that point generally alters the forecast.
Second. Another element always present in forecasting situations is uncertainty. If the manager were certain about the circumstances that will exist at a given time, the preparation of a forecast would be trivial.
Third. The third element, present to a variable degree in all the situations described, is the confidence of the person making the forecast on the information contained in historical data.
Forecasting Method Selection
• Factors
o The forecast context
o The relevance and availability of historical data
o The desired degree of accuracy
o The period of time to be forecast
o The cost-benefit analysis of the forecast
o The point in the life cycle of the product.
CLASSIFICATION OF FORECASTING MODELS
| Qualitative | ||
| Forecasting Methods | Quantitative | Time series analysis Causal models |
Qualitative Methods
Uses of these methods. Qualitative techniques are used when data is scarce, for example when a new product is introduced to the market.
These techniques use the judgment of the person and certain relationships to transform qualitative information into quantitative estimates.
Delphi method. It is used for long-term forecasts, new product sales forecasts, and technology forecasts.
Estimated time, more than two months.
Accuracy, fair to very good.
Market research. It is used to evaluate and test hypotheses about real markets.
Estimated time, more than three months.
Accuracy can be excellent, depending on the care that has been put into the work.
Consensus of a Panel. It has the same uses as the Delphi Method.
Estimated time, more than two weeks.
Accuracy, from low to regular.
Visionary Forecasts. It is used to make a prophecy of the future using personal intuition.
Estimated time, one week.
Accuracy, bad.
Historical Analogy. It is used for new products, based on the comparative analysis of the introduction and growth of similar products.
Estimated time, more than a month.
Accuracy, good to fair.
Quantitative methods
Time series analysis. Analysis consists of finding the pattern of the past and projecting it into the future.
Patterns of a time series:
• Horizontal or stationary
• Long-term trend
• Seasonal effect
• Cyclical effect
Projection methods. These methods try to find the total pattern of the data to project them into the future, and they are:
• Moving Averages
• Exponential Smoothing
• Box-Jenkins
Separation method. It is the one that separates the series into its components to identify the pattern of each component, and is called the Time Series Decomposition Method.
Causal Models
Regression Models
- Simple linear regression Multiple linear regression
Econometric Models. An econometric model is a system of interdependent regression equations that describes some sector of economic activities, sales, or profits.
Surveys of purchase intentions and anticipations. These surveys that are made to the public, determine:
to. Purchasing intentions for certain products.
b. They derive an index that measures the general sentiment about present and future consumption and estimate how these sentiments affect consumption habits. This approach to forecasting is more useful than other techniques for tracking demand and pinpointing danger points.
Input-output model. Analysis method that determines the interindustrial or interdepartmental flow of goods and services in an economy or in a company and its market. It shows flows of inputs that must occur to obtain certain outputs.
Download the original file
