- 作者: 楊健民; 劉立倫
- 作者服務機構: 政治大學資管系所; 國防管理學院會計學系
- 中文摘要: 人工智慧的範例學習法,在過去的研究上展現相當不錯的分類效果與預測績效。因此,本研究採用範例學習法,來預測企業經營績效;並根據預測的結果,形成不同的投資組合。研究結果發現,採用範例學習法所建構的決策樹,不僅可以有效的預測企業的經營績效,更在形成投資組合與獲得超常報酬上,展現了相當不錯的結果。而在範例學習法與Logit分析比較後亦可發現,範例學習法-強態模式所建構的模式,不僅在其模式的區別率、命中率上,顯示出較佳的預測結果;同時在投資組合的超常報酬上,亦較Logit分析的結果為佳。因此,我們認為,範例學習法可作為企業績效預測,與投資組合管理的另一種有效研究途徑。
- 英文摘要: Artificial intelligence-based rule-induction approach demonstrated good classification results andperformance prediction in previous research. In this study we adopted Learning-From-Example [LFE]approach to predict business performance and to establish investment protfolio. The finding suggestedthat the decision tree, which inducted from LFE approach, predict business performance effectively; andfurthermore, by using that predicting result to construct investment portforlio also secure good performancein protfolio management process. After comparing of LFE approach with traditional Logit approach, wefound that LIFE approach also prove to be more accurate in prediction process and to be better portfolioperformance. Therefore, we propose that LFE approach can be taken as an effectively alternative approachin business performance prediction and portfolio management.
- 中文關鍵字: 範例學習法; 經營績效; 資訊價值; 超常報酬
- 英文關鍵字: learning from examples; performance; information value; abnormal return