© 2000 John Petroff
|
|
© 2000 John Petroff |
H- Expert opinion, decision tree and simulation
This group of techniques are primarily used for forecasting with a probabilistic approach. In other words, the objective is to find the most likely value of what needs to be forecasted by evaluating the probability of all alternatives that can be reasonably expected. The first two methods, expert opinion and decision tree, are used extensively in all aspects of business decisions, even if decison makers do not formally identify their methods by these scientific names. They are more explicitly recognized in many judgmental economic forecasts outlined in Chpater 15 Section C-4.
Another technique consists in informing survey participants about partial results obtained from other participants. This encourages convergence toward a consensus forecast. To avoid stubborn outliers, the median rather than mean is chosen as the resulting forecast. This is called the Delphi process.
See review questions Q-5H1.1 through Q-5H1.3.
When a process involves a sequence of events (e.g. introduction of a new product in one selected store, then introduction in a region, then advertising campaign with promotional sale, and finally, introduction of the product nationwide), the questionnaire can ask for outcomes at each stage, and a conditional probability tree can be built, which is again used to calculate the mean of the distribution.
Judgmental forecasts have the advantage of reflecting implicitly in the opinions of experts all the most up to date and sensitive data available, and allowing a multitude of assumptions. Something that neither econometric models nor time-series analysis can do systematically. Yet the lack the apparent rigor of theoretically valid and explicitly stated assumptions are apparent drawbacks.
See review questions Q-5H2.1 and Q-5H2.2.
See research assignment R-5.12.
Simulations are widely used in operations research, for instance in studying alternative production processes. Simulations are also used in finance and economics to allow an analysis of the consequences of a range of potential events. For instance capital budgeting decisions can be studied by incorporating not exact values for variables but a range of values with a given probability distribution. In other words, it allows to insert uncertainty in financial decision. The technique is used to allow a system of equations representing the process under study to generate a large number of outcomes with all the possible values of studied events using the assigned probability distribution. The probability distribution of variables can be obtained from expert opinion or from past statistics. When past statistics are used for the simulation it is desirable that the volume of data be sufficient to obtain the frequency of various outcomes: the volume must be in hundreds - if not thousands - of observations. When the values of studied events are taken from a random distribution then the procedure is known as a Monte Carlo simulation. In this case, there is obviously no problem in running as many thousands of iterations as one wishes.
-- example --
See review questions Q-5H3.1 through Q-5H3.3.
See research assignment R-5.13.
| Previous: Linear_programming |
|
Next: Market research |