Many decisions made in operation management are based on forecasts. This is the basic principle used in matching supply to demand. A company’s accurate forecast enables it to take advantage of future opportunities and decrease on any impending risk (Stevenson 109). Forecasting is basically an anticipation of an event in the future. In business, entrepreneurs need to forecast or make a valid anticipation of what a customer will want to buy. In operations management, forecasting is a way of getting information on what a customer will demand in future. This is a way of aligning the supply according to demand in the market. It determines the capacity of demand on a certain good at a specified time in the future. For a good planning of production in a factory, the demand should be known so that the right number of staffing and equipment will be determined for use. Therefore, forecasting is an essential tool that is required during planning for any business activity (Stevenson 74).
There are two important aspects that are related to forecasting: the level of accuracy that a forecast provides and secondly is the anticipated level of demand. The anticipated level of a market demand can be attributed to a function of structural disparity, for instance seasonal variations or known trends. The capacity of forecasters to correctly model demand is referred to as forecast accuracy. Forecasting affects several areas in a business model such as in finance, accounting, management information systems, human resources, operations, service or product design and marketing.
Managers use forecasting in yield management for planning on when to give discounts in order to match capacity with demand. There are two main uses for demand which include planning the system and planning the use of the system. Planning the system involves making plans over a long range of time whereas planning the use of the system is usually short term decision making.
Forecasting has its own shortcoming despite the use of sophisticated computer programs and mathematical models. A high level blending of scientific skills and intuition is required in order to make accurate forecasts. There are different forms of making forecasting but all these forecasts have factors that are similar all through. Some of these common factors are that; forecasting techniques assume that factor that affected a system in the past will still affect it in the future, forecasting give inaccurate results depending on the prevailing conditions that may have been left out, when making forecasts for individual items, errors are greater as compared to a group of items, and finally short term forecasting tend to be more accurate than the long term (Stevenson 76).
There are several elements that are related with a good forecast. Forecast should be used in a timely manner for the necessary actions to be taken in preparation for the predictions given by the forecast. A good forecast should have its degree of accuracy stated and this accuracy should be high. A reliable forecast is desirable as it should be highly consistent. Forecast should be well understood as it should be presented in units that the users or interpreters of the forecast are familiar with. It is also necessary to put a forecast in writing as this information will be subject to evaluation once actual results are obtained. Forecast must be simple to implement and it should be cost effective as well. Accuracy in forecasting is very important in the supply chain network. The supply chain might have excesses at all levels of the supply system (Stevenson 77).
There are six fundamental steps that ought to be followed in order to develop a good forecast. Firstly, the reason for the forecast should be determined. The timeline for its use will provide the details of level of prediction that the forecast will have. This will be followed by an establishment of a time zone with the fact that the longer the time the less the accuracy of a forecast. Data that is appropriate for this forecast should be obtained and ascertained to ensure it is correct for the purpose of analysis. After this, an appropriate forecasting technique is then determined for the purpose of this forecast. The actual forecasting is then conducted. Finally, the forecast must be monitored to ensure that the predictions made are conforming to reality and if there are disparities then re-examination of the upcoming issues should be conducted (Stevenson 78).
A forecast should be accurate in order to be of use; therefore any forecast should be void of errors. However, it is difficult to make correct predictions in the real world due to the dynamic nature of prevailing conditions. Every forecast should provide a degree of deviation from the true values. During the determination of periodic forecasts, forecast errors should be monitored to ensure that they are within reasonable ranges. The accuracy of forecasting is a very important factor when choosing the right alternative among the many possible options. Historical errors of performance are the basis for making accurate forecasts.
Forecasting has two main approaches: qualitative and quantitative. For qualitative approach, subjective inputs are used which are usually in contravention of precise numerical depiction. Quantitative approach makes its forecasts by the projection of chronological data or through the creation of associative data that are focused on explanatory variables to make a prediction. Forecasting can further be classified into judgmental forecasts which are built on the analysis of subjective input gained form research surveys conducted on the various business stake holders. Time-series forecasts try to predict the future trends using past experiences. Associative models are based on mathematical equations which try to predict trends using explanatory variables.
Sometimes forecasts may solely depend on opinions and personal judgments to make a forecast. This especially happens when quick forecasting is necessary and there is not enough time to collect the required information to facilitate quantitative forecasting. During the introduction of new products into the market, there is usually no past history to make future predictions on. Therefore, during such times, forecast is done based on customer surveys, opinions of experts, opinions from the sales staff and executive opinions. The management level staff may convene a meeting and create a forecast. This approach is mostly viable when long range forecast are being made in new product development and planning. It’s advantageous since it brings together great management talent. However, the possibility of ones opinion prevailing over the rest is high which may result in the production of a forecast that may lack in accuracy.
Sales force always provides direct information from directly interacting with the customers. They may have information of the customer’s future plans. The drawback on using this information for predicting is that the sales person may not be able to tell between what the customer will do and what he would wish to do. This opinion may not be accurate too since sales persons may have there judgments affected by the business trends. For instance when sales are good for the sale person, they tend to be over optimistic and when the sales are low they become overly pessimistic thereby making poor judgments of the situation.
In his journal, “Choosing business forecasting methods” Melanie Hickens state that all business, small or big, relies on forecasting methods daily in their day to day running (2013). Forecasting is rather a competition between business competitors since whoever makes accurate forecasts takes the day. Accurate forecast gives a road map for how to make plans by avoiding risks and concentrating on the business opportunities. Melanie further goes ahead and gives eight points that he perceives to be the pillars for a good forecast. These includes information on availability of historical data, amount of time and money available for spending, the time the product has spent in the market, the kind of product or service being offered, whether the company is experiencing economic downturns, number of years of historical data available, how far ahead is the forecast determining and finally the importance of the focus to the company (Melanie 1).
Consumer surveys can provide information that may not be found elsewhere. The consumers are the final determinants of the demand of a good or service therefore their opinion is paramount. Conducting consumer survey needs a lot of knowledge and skill in order to be able to interpret survey result correctly. This may be an expensive undertaking and time consuming. Other approaches of making forecasts are by seeking for opinions from other managers from outside the organization. The Delphi method tries to create a consensus by issuing questionnaires to determine personal opinions from people who have the knowledge and ability to provide significant assistance (Stevenson 81).
Forecast can be made basing them on time series statistics that are made on a regular interval. This kind of focusing assumes that future trends will conform to those of the time past series. This method often produces satisfactory results although no attempts are usually made to pin point on the variables. The fundamental behavior of the series provides an analysis pattern of the time data series which may be grouped as seasonal variations, trends, variation around an average or irregular variations.
The naïve approach is a widely utilized approach of making forecast although it quite simple. It uses only one earlier data value of a time series to make predictions. It is commonly used around a stable trend of series which vary around a mean average. When a stable series is identified, the last data values are used as a basis for forecasting on the next possible trend. For instance, the demand for chocolate this festive season is equal to the demand in last year’s festive season.
Most data historically harbors a random variation which could be as a result of influence from factors that could be considered as relatively unimportant. Averaging techniques are used to smooth these variations in data. The available techniques for making these averages include; the moving average, weighted moving average and the exponential smoothing. Moving averages forecast uses data recently collected in generating a forecast. It is a way of overcoming the weakness of the naïve forecasting which traces actual data but has a lag of one period (Stevenson 83). The moving average helps in smoothing this data by incorporating more recent data thereby eliminating this difficulty.
The weighted moving average is closely related to the moving average only that more recent data are assigned more weight in the time series. A more sophisticated weighted averaging method is the exponential smoothing although it’s equally easy to understand and utilize. Every fresh forecast is related to the last forecast with an additional proportion of the variation between the real value of the series at that time and the forecast.
Other approaches for forecasting may include focus forecasting and diffusion modeling. Focus forecasting uses an analysis of the best current performance. This method combines several methods of forecasting, for instance weighted average, moving average and exponential average, to the recent historical data after the elimination of variations. The forecast for the next month will be best on the method that was highly accurate. For diffusion modeling, during the introduction of new products in the market, usually there is lack of historical records of sale for forecasting. Instead other products sales are used to make a prediction. This model takes into account the range of mass media, word of mouth and the market potential.
Techniques for trends deals with the development of equations that is highly suited to make a description of trends. Trend-adjusted exponential smoothing is used only during instances when data variation is around an average or has gradual changes. Therefore it’s only used when time series portrays a linear trend. Seasonality variation can be described as annual variations that move upwards or downwards repeatedly and can therefore be related to a recurrent event. The associative forecasting technique depends on the classification of connected variables which can be utilized in predicting variables that are of concern (Stevenson 98). An example of this is that the sale of an orange may be directly linked to the price of an alternative for example an apple. This technique uses the simple linear regression in making its predictions which basically involves linear connections of two variables.
Randolf Saint, from his journal on business forecasting gives an insight of how to use historical data and regression analysis in making business forecasts (2011). Most forecasters, use past performances of businesses as a building block for making future predictions on the state of the business affairs in future. Regression analyses utilize independent and the dependent variables statistically to find a relationship of the two thus making a prediction. Historical data that may be used include the clients’ invoices, company’s financial statement and other documentation that are deemed to be of importance in making predictions. Almost all business fields can utilize regression analysis. In this model, the dependent variable could be a company sales and the independent is the interest rates. The use of historical data for forecasting has its limitations. The historical data obtained could have been as a response to a series of events that may not be repetitive in future (Randolf 1).
Forecast must be monitored to check on the errors that may have been overlooked during the forecasting and make the necessary changes. Most forecasts are done at time intervals that are regular. Tracking of these errors throughout the forecast period gives an insight of the general performance of a forecast. The probable sources of errors in forecasting include: inadequate models due to omissions, introduction of variables that a model cannot deal with or appearance of a new variable; severe natural phenomenon may cause irregular variations; incorrect use of the forecast technique; and inherent randomness that are found in data after all causes of variations are eliminated.
Work Cited
William, Stevenson. Operation Management. 11th Ed. New York: McGraw, 2011.
Kasey, Wehrun. “Business Forecasting in a Crazy, Mixed-up World.” Inc. magazine 1 April 2009. Print.
Randolf, Saint. Business Forecasting Using Historical Data and Regression Analysis. Houston Chronicle, 1 January 2011. Web. 16 June 2013.
Melanie, Hicken. 8 Questions to Ask When Choosing Business Forecasting Method. Open Forum, 30th January 2012. Web. 16 June 2013.
Wilson, Holton and Barry, Keating. Business Forecasting. New York: McGraw-Hill, 1998.
