Forecasting Project Performance Using A Cross-Entropy Neural Network

カテゴリ: 全国大会

論文No: 3-044

グループ名: 【全国大会】平成22年電気学会全国大会論文集

発行日: 2010/03/05

タイトル(英語): Forecasting Project Performance Using A Cross-Entropy Neural Network

著者名: ングブルアリック (北海道大学),本間 利久(北海道大学),大和 諸祖(NECコーポーレッション)

著者名(英語): Alick Nguvulu(Hokkaido University),Toshihisa Honma(Hokkaido University),Shoso Yamato(NEC Corporation)

要約(日本語): This paper forecasts project performance assessment indicator (PAI) using a neural network implementing a cross entropy error function. Four neural network simulations were carried out using past measured PAI and virtual PAI data. Virtual PAI data were generated by the Simple Mean, 2nd Lagrange interpolation, and 2nd Newton interpolation methods. The neural network model performance was assessed by the relative absolute error (RAE) and mean absolute percent error (MAPE). Predictive performances in the order of 0.290 (2.100%), 0.237 (1.854%), 0.102 (1.745%), and 0.100 (1.740%) based on RAE (MAPE) were achieved by the model on the four respective simulations.

原稿種別: 日本語

PDFファイルサイズ: 2,361 Kバイト