A decision tree is a graphical technique to document and
evaluate possible outcome of multi-stage decisions taking into consideration
probabilities of different outcomes of a decisions taken at any stage. The decisions and
their possible outcomes are represented in a diagrams that branches out at each stage of
the multi-stage decision. Thus the diagram starts with a single decision which is like
the stem of the tree. Different outcomes of this decision with their probability of
occurrence are represented as branches emanating from the initial decision. For each of
these possible outcomes managers will be required to decide on further course of action.
Thus there will be further sub-branches of the decision tree from each of the first
level branch. This kind of branching of decisions and their representation on the
decision tree can be for several stages.
The representation
of decision tree in this way enables managers to identify all the possible outcome in a
multilevel decision situation, and evaluate the impact or attractiveness of each
possible final outcome along with the probability of its occurrence. This enables
managers to choose alternatives that is likely to give best
results.
For example, a company considering expansion in
manufacturing capacity can consider several alternative in terms of the size of new
plants to be added. Having added the manufacturing capacity the actual sale and profit
will depend on the future increase in market demand. Depending on the percentage actual
increase in market demand, for each of the capacity increase scenario, the company will
again have to take decision on actual quantities to be manufactured and prices to be
charged. There will be several possible decisions for each of the plant capacity chosen
at first level decision. Also for each of this decision the actual sale will depend
on demand elasticities, which is uncertain. Thus the company can estimate probabilities
of different demand base levels and demand elasticities to assess outcomes of the next
level decisions.
No comments:
Post a Comment