Integrated multiple-attribute decision making and kernel-based mechanism for risk analysis and evaluation

Due to the radical changes in the global economy and internationalization of markets around the world, local corporates are encountering much more severe challenges and uncertainties than ever before. Hence, how to reliably and effectively evaluate whether corporates will exhibit substantial troubles/difficulties in the near future turns out to be an attractive investigative issue. This study introduces a fusion mechanism that gives decision makers a comprehensive description on a corporate's operation status so as to prevent a biased judgment from occurring. The introduced mechanism consists of three main procedures: (1) Performance rank determination through the integration of balanced scorecards (BSC) and two-level DEA with a weighting adjusted strategy; (2) Forecasting model construction by combining core vector machine (CVM) with the support vectors (SV)-based online learning strategy; and (3) Knowledge extraction by a rule-based algorithm. The experimental results show that the introduced fusion mechanism (i.e., TDMCR) reduces unnecessary information, gives a more overarching description, satisfactorily predicts a corporate's operation status, and provides intuitive decision logics for market participators to adjust their investment portfolios for maximizing their profit margins under an anticipated risk level. Examined by real cases, the introduced fusion mechanism is a promising alternative for corporate operating performance forecasting.