Comparing and Contrasting Forecasting Methods

Comparing and Contrasting Forecasting Methods

Companies use forecasting to help decide how to best spend funds for the next year, to predict if expansion is needed, to plan for how much of each product to produce within a certain period of time, and other decisions that effect the company’s future plans. Qualitative, time series analysis, causal relationships, and simulation are the four basic types of forecasting (Chase, Jacobs, & Aquilano, 2006). The forecasting methods that will be compared and contrasted within this paper are the Delphi method (which is an example of qualitative), time series analysis, seasonal, and causal relationship forecasting.

The Delphi method is a judgmental forecasting method, which uses the evaluation of several experts within the field that is being analyzed to forecast company sales. The process starts by contacting several experts and asking them to participate in the study of the company. When multiple experts agree to join in the research, the company’s material is sent to each specialist to evaluate, and are asked to send the findings and materials back to the company with suggestions and predictions attached. A company coordinator studies all of the information and forecasts, has key company executives make additional comments and reflections, then asks the experts if they want to make any changes to the information they provided. This will happen several times until all parties involved reach a consensus. “Forecasts based on a group of forecasts are better than forecasts of a single forecaster, particularly where no formal forecasting process exists…” (Ahamad & Ismael, 2003, p. 22). For the Delphi method to be used effectively, it is important to keep the identity of the experts used undisclosed, so those that are providing information to the company know it will not be used against any participant at a later time and make the experts feel comfortable to be honest in the review provided. The Delphi method is used when little or no historical data is available, making this method very useful for new companies, but not necessarily the best method if historical data is available for comparison.

“Time Series Forecasting (TSF), the forecast of a chronologically ordered variable, corporals an important tool to model complex systems, where the goal is to predict the system’s behavior and not how it works,” (Cortez, Rocha, & Neves, 2004, p. 415). Several different forecasting methods exist within the time series category, including simple moving average, weighted moving average and simple exponential smoothing, exponential smoothing with trend, and linear regression. To decide which of these methods are best to use for forecasting, time horizon to forecast, data availability, accuracy required, size of forecasting budget, and availability of qualified personnel should be taken into consideration (Chase, Jacobs, & Aquilano, 2006). Time series analysis uses past data to try to predict future events. Therefore, this type of forecasting does not work for newly formed companies, but this type of forecasting method does work for companies that have many years in the industry.

Seasonal forecasting is a form of forecasting that also uses historical data to predict sales within a certain season, such as spring for flowers, fall for harvest, or summer for travel. It can also be used in trying to predict certain weather conditions like hurricanes and tornados, or for companies that produce office supplies the beginning of the school year might be considered a season. “Seasonality is so strong in many industries that losses routinely occur in the off-season. It causes elementary and secondary textbook businesses, for example, to incur operating losses in the first two fiscal quarters,” (Radas & Shugan, 1998, p. 296). After many years of collecting information, future sales can be predicted by quarter, season, or peak of sales that have been established.

Causal relationship happens when one independent variable causes an occurrence of another independent variable. For example, when snow causes the sale of snow shovels and ice scrapers to increase. If it is known ahead of time that it will snow, the increase of sales of snow shovels and ice scrapers can be predicted. “The first step in causal relationship forecasting is to find those occurrences that are really the causes. Often leading indicators are not causal relationships, but in some indirect way, they may suggest that some other things might happen,” (Chase, Jacobs, & Aquilano, 2006).When using causal relationship forecasting, finding each and every relationship between the unknown and the product the business is selling could generate added income. Causal relationship forecasting can be used in conjunction with other methods of forecasting to provide an even more precise prediction.

In