獲利能力就是企業資金增值的能力， 獲利能力分析是企業為產生利得與所投入可運用資產、資本及資源的關係。一般獲利能力之分析包括營業收入、營業成本、營業損益、稅後損益、各種損益項目之比率分析，此外也經常對各種投資報酬率作分析例如稅後淨利、每股盈餘、資產報酬率、股東權益報酬率等，來衡量一家企業整體之經營績效。
本研究針對台灣上市櫃半導體、光電、電子零阻件、通信網路、鋼鐵、塑膠、化工等7大產業104家公司，探討其在本業獲利上從2003到2016年營業過程中1074筆公開資料，建立獲利能力預測模型。先以逐步迴歸分析及相關係數將變數做一篩選，瞭解有顯著影響的自變數，並以倒傳遞類神經網路模式預測企業投入與獲利能力之關係。最後再另選三大產業水泥 (台泥)、橡膠 (建大)、食品工業 (聯華食)實際驗證所建立的模型之有效性。再就各項輸入變數對輸出變數進行敏感度分析，以判定輸入變數對輸出變數的影響程度。
　　　研究結果顯示，以7大產業104家公司，879筆訓練資料及195筆測試資料進行神經網路監督式學習，對不同產業類別有不錯的預估結果，企業的營業毛利與所投入的研究發展費用、推銷費用、管理費用等對稅後淨利與資產報酬率有顯著的影響。

關鍵詞：獲利能力、多元逐步迴歸分析、相關係數、倒傳遞類神經網路
　　　　　敏感度分析

　　Profitability is the ability of a company to increase its financial value. Profitability analysis is the relationship between the company's profits and the available assets, capital and resources. The general profitability analysis includes operating income, operating costs, operating profit and loss, post-tax profit and loss, and the ratio analysis of various profit and loss items. In addition, it often analyzes various investment returns, such as net profit after tax, earnings per share, return on assets, the rate of return on equity, etc., is a measure of the overall business performance of a company.
　　In this study, using data to establish profitability prediction model by 104 companies in 7 industries, including Taiwan's listed semiconductors, optoelectronics, electronic zero-resistance devices, communications networks, steel, plastics, and chemical industries, were investigated and 1074 of them were disclosed in the process of profitability in the industry from 2003 to 2016. Firstly, a multiple stepwise regression analysis and correlation coefficient were used to make a selection of variables, to understand the variables that had significant influence, and to predict the relationship between corporate investment and profitability using the back-propagation neural network model. Finally, the validity of the model established by the actual verification of the three major industrial cement, rubber, and the food industry was selected. Then, the sensitivity of the output variables is analyzed on each input variable to determine the degree of influence of the input variables on the output variables.
　　　The results of this study showed that 879 training materials and 195 test data from 104 companies in 7 major industries were used to perform neural network supervised learning. There were good prediction results for different industry categories, and the company's operating margins and research and development were invested. Expenses, promotional fees, and management fees had a significant impact on the after-tax net profit and return on assets.
Keywords: Profitability, Multiple stepwise regression, Correlation coefficient,
Back-propagation neural network, Sensitivity analysis

[1] MBA 智庫百科 (2014)，財務分析中的五力分析法
[2] 蘇朝墩 (2013)，品質工程：線外方法與應用，前程文化事業有限公司。
[3] 蘇朝墩 (2009)，六標準差，前程文化事業有限公司。
[4] 蘇朝墩 (2017)，Introduction to Neural Networks 上課教材。
[5] 桑慧敏 (2007)，機率與推論統計原理，美商麥格羅希爾國際股份有限公司。
[6] 陳智彥 (2009)，企業評價與內部管理績效關係之研究,
[7] 葉怡成 (2000)，類神經網路模式應用與實作，儒林圖書公司。
[8] 陳澤義、陳啟斌 (2014)，企業診斷與績效評估，華泰文化。
[9] 謝劍平 (2015)，財務報表分析概論，智勝文化事業有限公司。
[10] 李惠妍、吳宗正、溫敏杰 (2006)，迴歸模式與類神經網路在台股指數期貨預測之研究，經營管理論叢, Vol.2, No.1, 第83-99頁。
[11] 彭光正 (2010)，以迴歸樹與逐步迴歸建構台灣股市多因子選股，中華大學資
訊管理，碩士論文。
[12] 羅煌木 (2010)，以類神經網路預測垃圾添加灰燼之氣體產量之研究，朝陽科技大學環境工程與管理系。
[13] 孔祥竹 (2007)，應用類神經網路減少TFT-LCD產品測試項目之研究，國立交通大學工業工程所，碩士論文。
[14] 郭意昇 (2006)，企業經營失敗原因之探討，中正大學企管所，碩士論文。
[15] 陳順宇 (2005)，多變量分析，華泰文化。

[16] 薄喬萍 (1979)，應用MINITAB建立都市道路交通迴歸模式之研究，淡江
大學管理研究所，碩士論文。
[17] Ching-Hua Liou, Yaw-Shyan Huang, Ren-Fen Lee (2010), The Study for Business Profit Factor-Electronic Industry as an Example/Journal of Commercial Modernization, Vo1.5, No.3,1-18.
[18] Hiroshi Saito, Ken Takiyama,, and Masato Okada (2015), Estimation of State Transition Probabilities: A Neural Network Model, Journal of the Physical Society of Japan.84
[19] Meijer, S.C.F. (2004), Theoretical and practical aspects of modeling activated sludge processes, Delft University of Technology, The Netherlands, 47-48.
[20] Van Veldhuizen H.M., Van Loosdrecht M.C.M. and Heijnen J.J. (1999), Modeling biological phosphorus and nitrogen removal in a full scale activated sludge process. Wat.Res.,33 (16), 3459-3468.
[21] Chang,C.H., Yang.C.,Tung,Y.K. (1993), Sensitivity and uncertainty analysis of a sediment transport model: a global approach. Journal Stochastic Hydrology and Hydraulic, 7(4), 299-314.
[22] JingTao Yao,Chew Lim Tan (2001), Guidelines for Financial Forecasting with Neural Networks.
[23] Iebeling Kaastra,Milton Boyd (1996), Designing a neural network for forecasting financial and economic time series, Neurocomputing 10, 215-236.