Pruned decision tree 

Contexts in source publication

Context 1

... classified by a decision tree into insecure (at least one of the 2 lines overloaded) and secure situations. The outages on 150.10 and 150.9 are called parallel and the 3 remaining outages other. The decision tree (6728 objects in the Learning Set, 1121 in the Pruning Set and 1121 in the Test Set, the er- ror probability is 4.728%) represented on Fig. 5 confirms the experience of the power system engineers. The tree is read top-down, and each decision node is represented by a box showing in its upper part the percentage of secure and insecure scenarios in the training and in its lower part the test set error rate. The tree of Fig. 5 shows that a conges- tion on one of those 2 lines ...

Context 2

... the Test Set, the er- ror probability is 4.728%) represented on Fig. 5 confirms the experience of the power system engineers. The tree is read top-down, and each decision node is represented by a box showing in its upper part the percentage of secure and insecure scenarios in the training and in its lower part the test set error rate. The tree of Fig. 5 shows that a conges- tion on one of those 2 lines can be fixed either by stopping a generating unit at sub-station E or by switching on the parallel line or by decreasing the generation around sub- station R. We mention the fact that a reinforcement has already been identified to fix the problem, but it was not considered in this ...