Stay Ahead with the Power of Upskilling - Invest in Yourself! 
Forecasting Models – Production and Operations Management
Quantitative forecasting models are used to forecast future data as a function of past data.Examples of quantitative forecasting methods are last period demand, simple and weighted N-Period moving averages, simple exponential smoothing, poisson process model based forecasting and multiplicative seasonal indexes.
OTHER TIME-SERIES METHODS
A modified version of the weighted moving average technique is the Exponential smoothing Method. The simplest way in which it could be expressed is as follows:
Suppose we have an old forecast made for the present period as 100, the actual sales observed in the present period is let us say, 90. Then the new forecast (for the next period) by the exponential smoothing method is obtained by giving weightage to the old forecast and the actual sales. Let us say the weightage given to the current actual sales is 0.2 and the weightage given to the old forecast is 0.8.
Now, the new foreorecast = (100 x 0.8) + (90 x 0.2) = 98
The advantage of this method over the moving average method is that, one needs to have only two figures: one for the old forecast and another for the actual sales observation. It is not necessary to store the data on a number of past periods.
Trend Correction
The correction to the trend component combined with the cyclical component
Correction for Seasonality
Seasonality influence is primarily due to customer’s purchase policy preferences, periodic government policies and climatic changes. Such steep rise or absence of demand during various time periods is referred to as peaks/ valleys and is compared with the general average demand during the non-seasonal periods.
Procedure for Using Exponential Smoothing
1. recording of the demand in past year usually represented in monthly demands .
2. If significant seasonal variation is observed, then a base series is formed. This series could be the demand for the last year repeated verbatim, or if the seasonal periods themselves are slightly fluctuating a cantered moving average is found for all the past data.
3. The base series is computed and current month’s demand is compared with base series demand to find the demand ratio.
4. Now, demand ratio calculate dearlier, is taken as forecast for next period and is also called as forecast ratio.
5. A forecast for the next month’s demand ratio is made by noting the previous month’s forecast ratio, the alpha factor and the current month’s demand ratio.
6. The forecast ratio now is corrected for the trend component by first finding the trend factor from the previous observations of the demand ratio. It is the corrected forecast ratio to be used for the next period in the ocmpany.
7. The corrected forecast ratio has now been rectified for any random component as well as for trend and cyclical factors. The seasonality is already included due to usage of this ratio. Therefore, the next step is to get the forecast of the demand by multiplying the forecasted demand ratio by the base series demand observed for the corresponding month.
LIMITATION OF TIME-SERIES METHOD
One needs to consider these models in terms of their relevance and peculiar drawbacks. The methods we have considered so far, viz. moving averages, exponential smoothing, etc. can be grouped under the category of time series models. In actuality, the demand may vary due to various market and other external and internal factors. The time-series models club together a whole lot of possibilities or reasons for variations in demand in terms of one factor, that is, time.
After making the forecast, it needs to be monitored for errors or deviations from the actuals observed. This is necessary in order to make any modifications in the originally assumed forecasting model. There are two measures of deviations.
FORECAST ERROR MONITORING
Mean Absolute Deviation (MAD)
‘Absolute’ here means that the plus or minus signs are ignored; and ‘deviation’ refers to the difference between the forecast and the actual.
Running Sum of Forecast Errors (RSFE)
This is the algebraic sum of the forecasting errors, which means the negative and positive signs are given
their due significance.
The RSFE is a catenation to determine whether or not the forecast has any positive or negative bias. A good forecast should have approximately as much positive as negative deviation. MAD indicates the volume or amplitude’ of the deviation from the actual. Both the ‘bias’ as well as the ‘amplitude’ of the forecast errors are important. Therefore, it is important to monitor both the tracking signal and MAD for any modifications to be made in the original forecasting model.
INPUT-OUTPUT ANALYSIS
Many input-output type of models are also useful in forecasting. Input-output analysis takes into consideration the interdependence of the different sectors in the economy. An input to a sector is an output from another sector. For instance, an input from the steel sector might give rise to an output from the electricity sector, which in itself is an input to the steel sector. There are many such seemingly cyclical relations within the various sectors of the economy.
Such analysis is very useful because it takes into account all the intricate relationships in the economy. But still each technique has its own drawback. The drawback here is that the utility is restricted to economic analysis, not considering the other business, governmental, technological, and internal factors. It is a limited but useful analysis. The analysis need not be restricted to a macro-level, speaking only in term of the steel sector and electricity sector. It may be more ‘micro’, by considering the inputs and outputs within a general product group in the total economy. Such analysis has been done in practice and has been found to be useful.
DELPHI METHOD OF OPINION GATHERING AND ANALYSE
For a new product or service, the Delphi method, consisting of systematic gathering, analysis, and convergence of experts’ opinions, is very useful. This method occupies an important place in ‘technology forecasting’ which in itself is a very vital aspect of forecasting, living as we are in an age of rapidly changing technology. A note on Technology Forecasting is furnished at the end of this chapter.
RANGE OF FORECAST
A forecast may be in terms of ranges rather than exact figures. Although one may hit upon the exact figure after calculations, assuming various plausible changes in the environmental conditions, it is better to give a range for the forecast. This helps in many ways. It shows what the highest figure to be expected is and what is the lowest.
FORECASTING AND THE INDIAN SCENARIO
The Indian economy is increasingly getting the characteristics of a buyer(s) market. The Indian businessman, therefore, has to be very alert about the rumblings in the gangways. Forecasting models, such as the causal models can now be used to forecast the effect of concessions in corporate tax, customs duty, excise, and other areas. Opinion-based methods such as Delphi techniques and Consumer Behavioural Surveys have increasing relevance. Monopoly or oligopoly does not need forecasts; it is the competition that needs the forecasts. Indian industries and businesses are waking up to the fact that it is now a different game. They know that if they do not follow appropriate management basics such as forecasting, they risk the danger of being marginalised for a long time to come.
Production professionals, operations executives, managers, senior executives can use the below links to be updated on Production and Operations Management
- Practice Test for Production and Operations Management
- Certification in Production and Operations Management
- Blog posts on Production and Operations Management
Back to Production and Operations Management Tutorial Main Page
