結合資料採礦技術與類神經網路預測台灣IC設計公司之企業績效：以平衡計分卡為例

Abstract

台灣IC設計公司為了滿足企業診斷和績效預測，本研究整合具備智能回歸技術的變數選取方法來量化平衡計分卡（BSC）的概念。特別是，能有效解決以下關鍵問題：（1）績效指標均影響一個特定的結果能更具有代表性？ （2）通過調整所確定的指標，如何提高到預先設定的程度（即增加10％）的結果？ （3）如何披露相關的關鍵績效指標（KPI）和具體的成果之間的因果關係嗎？在此背景下，平衡計分卡（BSC）的四個維度，如財務、顧客、組織及營運，構建以應對上述問題。儘管許多研究已經提出來解決這個問題，但是，他們大多是完全依賴於專家的評估。同時，多餘的變量或冗餘的變數往往禁止決策者從圖紙管理的影響。因此，在一個數據驅動的方法，本文提出，以協助台灣IC設計公司進行業績預測和情境模擬，以及對結果的認識KPI和因果影響的重要性權重。

In order to fulfill business diagnosis and performance forecasting for Taiwanese IC design companies, this study integrates feature-selection methods with intelligent regression techniques to quantify the concept of the balanced scorecard (BSC). In particular, the following critical issues are well addressed: (1) Which performance indicators are more representative to influence a specific outcome? (2) How to improve an outcome to the preset degree (i.e. increasing 10%) by adjusting those identified indicators? (3) How to disclose the underlying causalities between key performance indicators (KPIs) and specific outcomes? In this context, a balanced scorecard (BSC) associated with four dimensions, such as financial, customer, internal process, and learning and learning and growth perspectives, is constructed to respond to the aforementioned issues. Despite numerous studies have been presented to tackle this problem, however, most of them are fully reliant on experts’ assessments. Meanwhile, superfluous variables or redundant features often prohibit decision-makers from drawing managerial implications. Therefore, in terms of a data-driven approach, this paper is presented to assist Taiwanese IC design companies in carrying out performance forecasting and scenario simulation, as well as recognizing the importance weights of KPIs and their causal impacts on outcomes.