Short essay on selection of optimal model by using experimentation

Selection of optimal model by using experimentation

Before selecting the better model, the available models are narrowed down to five or six or may be even less than that. The models selected must not be either custom made or mathematically complex and complicated. Selection of the appropriate models is dependent on the nature and complexity of the problem under consideration and investigation.

Two procedures are well-known in OR literature for deriving an optimal solution from a model, viz. analytic and numerical. Numerical procedures employ computers and are concerned with trying various values for the control variables in the models, comparing the results obtained and selecting those control variables that yield an optimum solution. On the other hand, analytic procedures employ mathematical deduction – such as simple algebra and calculus.

Implementation and verification of the Model:

The last but one step in OR are the verification through implementation of the model. In order to verify the optimum model, it must be translated into a set of operating procedures capable of being understood and applied to the personnel who will ultimately responsible for their use.

To test the efficiency of the model selected, of course, dual or parallel operations may be necessary. While implementing the model, the OR specialists should convent the mathematical language into simple one so that managers can understand the meaning and significance of a particular solution.

Establishment of Proper Control:

OR models require continuous feedback and monitoring. Such monitoring provides a means for modifying die system as and when the external and internal conditions demand change. OR is continuous and the teams should modify the models depending on the significance of changes in business and economic environment.

A solution derived from a model remains and optimum as long as the variables retain their original significance and relationships. OR experts feel that the solution goes out of order when one (or more) of the relationships between variables has (had) changed phenomenally. Therefore, when a firm is operating in a dynamic as opposed to stable environment, continuous monitoring is necessary to establish control over the models that have been implemented.