What are Reinforced Decision Trees

Decision tree

Decision theory

1. term: Form of representation of multi-level decisions. The decision tree is obtained from an expansion of the status tree by including the available alternative actions (actions) in addition to the expected environmental statuses in the individual points in time.

2. presentation: In a decision tree, a distinction is made between rectangular and round branch nodes. Decision nodes are rectangular (the decision maker has to make a decision), state nodes are round (an environmental state occurs). In the example in the figure below, decisions are made in t = 0 and t = 1. The possible environmental developments after the first decision in t = 0 (choice of one of the alternatives A1 or A2) are S.1 and S.2. After one of the states in t = 1 has occurred, a further decision is made, e.g. after A's choice2 in t = 0 and occurrence of the state S1 between A7 and A8 to choose.

For reasons of clarity, no probabilities of occurrence for the environmental conditions have been included in the decision tree in the figure. If they are drawn in the tree (on the branches leading out of the state nodes), they represent transition probabilities, i.e. conditional probabilities for the occurrence of an environmental state under the condition that a certain state has occurred beforehand. A chain of alternative courses of action (e.g. A.1, A6) forms a strategy which, together with an environmental development (e.g. state S2, S.3) leads to a certain result [x (A1, A6, S.2, S.3)].

3. application: With the help of a decision tree, multi-level decisions can be represented and solved according to the principle of backward induction (roll-back procedure); see flexible planning. In the example in the figure, the decision maker first determines the optimal alternative for all four decision nodes in t = 1. Only then does it determine the optimal alternative for t = 0, whereby when choosing between A1 and A2 the optimal follow-up decisions are taken into account. The decision tree allows the decision-making situation to be fully mapped, but is only practical for very simple decision-making problems.

Industrial and organizational psychology

Procedure that goes back to Vroom and Yetton, according to which the degree of participation of the person led in the decision-making process of the leader in the sense of the situation theories of the leadership depends, among other things, on the degree of task complexity, the leader's level of information, the need for acceptance of the decision among those led, on the homogeneity of the objectives of Leadership and lead and the degree of conflict within the working group.

Evaluation: The model tends to be empirically confirmed and at the same time specifies conditions for the occurrence of synergies.