1-115 予測・診断モデル構造決定のための決定木適用の検討

概要

論文の詳細を見る
Multiple regression models and artificial neural networks have been applied for building various forecasting and diagnosis systems. However, it is difficult to obtain the accurate forecasting and diagnosis results by using only one model of neural network and multiple regression model. This problem can be solved with plural models. However, it is very difficult to decide the configuration and input factors for building appropriate plural models. This paper proposes a forecasting and diagnosis model construction method using decision tree methods. The proposed method adopted the statistical hypothesis test for the evaluation of generating decision tree. The decision tree is generated according to the characteristics of data distribution. Consequently, it is very easy to decide the configuration and input factors for building appropriate plural models. The effectiveness of the proposed method is demonstrated by applying a peak load forecasting model in power systems.
一般社団法人日本機械学会の論文
2002-11-14