Recently, the classification study is accelerated, especially
in machine learning expertise. Although the decision tree was
still recommended as a classification tool in diagnosing electric
power apparatus because of the property having the visible if-then
rule, the recent development in classification methods, especially
those using the ensemble methods, suggests us to apply these
methods to condition diagnosis area. In this paper, we report
that the new ensemble methods show extremely high accuracy in
classification of the electric power apparatus diagnosis, although
rule visibility is sacrificed.
|
|
|
|
|
|
Condition diagnosis, classification, decision
tree, diagnosis accuracy, misclassification rate, ensemble
methods, box-plot.
|
|
|