本論文之目的乃在為我國的社會福利支出水準找到一個定位，並預為我規劃社會福利 支出時的參考。 過去數十年，我國在「經濟成長」之目標主導下，對於「基本需要措施」(Basic Ne- eds Approach) 及講求社會公平之重分配功能則嫌忽略。試然，過去我們因受制於有 限財力及經濟障礙，對於社會福利方案的推動雖有相當的困難，但今日當這些受制的 因素逐漸地解除時，我們的社會福利水準是否已經有所提升？我們距離福利國家的目 標還有多遠？與各個不同開發程度的國家相較，我們的成績又如何？又在各個範圍不 同的社會福利定義下，我們的表現如何？這些問題，我們在論文中，都將一一予以回 答。 本論文由基本的社會福利支出模型建立為出發，並佐以實證分析，以找出影響各個不 同經濟發展程度國家之社會福利支出的因素，並據以計算各類指數，試圖以國際比較 的方式來找尋出各國理想與實際支出差異的情形，並希望藉此能回答前述的問題。同 時，我們也根據前述的迴歸結果，對我國未來的社會福利支出水準作預估。 所謂知已知彼，百戰百勝，由這個國際比較的方式，或可收他山之石以功錯，更希望 能預為我規劃未來社會福利支出時的參考。

平常對客戶有作經營績效分析，但未能從中去作客戶篩選之具體依據。期能透過此研究來幫助自己對客戶的篩選，找出可信度較高的方法來篩選客戶，並能從中預估其獲利。 　　在過去碩博士論文對於客戶獲利預估之探討，並沒有相關之論文能在數字上有所琢磨。倒是對於獲利模式或模型有一些研究的論文，而其著重在行銷及策略上，還未到經營分析的深入層面，或許因研究者較沒有企業實際運作的資料可用，就算有資料可用，其資料的足夠性如時間長度可能也不夠長因而無法作探討。正好欲研究的個案公司取得資料容易而且營運的時間也夠長夠穩定，可以針對客戶獲利預估方面的論文研究作更深入之探討。 　　所以本研究將會利用現有實際運作資料來搭配統計分析工具作獲利預估的相關性研究，由於是前面沒有人做過類似之研究探討，故可借此研究探討之發現，能對後續有此相關研究興趣者提供一些參考，而能做更多更深入的研究。 　　本論文研究探討的產業是伺服器產業，基本上此產業的主要客戶可分為下列四大族群，分別為品牌業者、網路業者、系統整合公司及電信業者。 　　而研究的個案公司是ODM伺服器代工廠，目前對這些客戶群都有業務往來。近來台灣的伺服器產業大都侷限在ODM代工角色，其貢獻全球40%以上的伺服器系統產量，這還未包括準系統及主機板的出貨，因為有些客戶要求在海外進行最後的系統組裝，或者直接將主機板放到已規劃好的資料中心內的硬體機架上。若加上這些準系統及主機板將超過90%以上的產品出自於台灣的代工廠。 由於日益增加的客戶愈來愈多，故如何去做客戶篩選，將是各家廠商所要考量的重點，尤其在人才有限的情況下，更因挑出有獲利的客戶來合作。而個案公司伺服器部門在過去五年每月的實際運作資料，將可以做為客戶篩選探討的資料依據。本研究將以每個客戶的獲利做為因變項，而影響獲利較顯著的投入人力及毛利貢獻為自變項，以及客戶特性為虛擬變項，利用多元迴歸分析找出客戶的獲利模型，然後將研究分析的結果與實際營運狀況作對照；接著運用迴歸分析所得到的模型作客戶未來獲利之預測；最後再探討如何建立既有客戶與新客戶的篩選標準。 經本研究後發現當選擇客戶的家數不同所產生的迴歸模型也不同，但主要的客戶家數一樣的話，其所得的結果對於獲利方向而言會很相近，但對於獲利預測值而言，有些客戶會很相近，而有些客戶就相差較大。故最後可歸納如下： 1. 迴歸模型可提供獲利的大概方向，但無法得到獲利預測的精準值。 2. 迴歸模型在一段時間後，須重新檢視其適用性或者加入後續的客戶經營績效重新更換新的迴歸模型作為新的判斷之用，才能更接近目前的營運狀況。 關鍵詞： 客戶篩選 獲利預估 迴歸分析 / In order to evaluate cost performance, companies generally analyze individual customers’ business performances to gasp a main idea. However, most companies are unable to screen out specific customers afterward based on the results. This study hopes to provide companies a way to screen clients with higher credibility and estimate likely profits generated from different clients. Many Master degree dissertations in the past have touched topics related to evaluating customers’ profitability. Yet, there are not many papers defining figures throughout the evaluating process. Many ongoing research papers are in view of constructing profit models that emphasis on evaluating marketing and strategy analysis, while papers on in-depth customers’ business performance are comparably less and more difficult to conduct because actual information on business operation arerather difficult to obtain, and even with information at hand, the adequacy of the data must be taken into concern. Fortunately, I was able to acquire the relative data on appointed companieswithin a specific time interval, allowing me to study further into evaluating customers’ profitability. Hence, this study will be use existing data along with statistical analyzing tools to estimate return profits. This thesis investigates the server industry, including respectively four groups, the brand owner, network operators, system integrators and carriers- the main customers of this industry. The case studied in the thesis is an ODM server manufacturing foundry, which currently hasbusiness interaction with all four groups of customers. Taiwan's server industry is mostly confined to playing a role in ODM or OEM, whereas its contribution of server systems builds up to more than 40% of the production worldwide, excluding the barebones and motherboard components, where some customers require the final assembly to take place overseas, or to directly embed the motherboardinto the hardware rack of a specific data base.Accumulatingthese productions as well, Taiwan produces over 90% of the world supplies. With a growing set of customers,how to select customers now becomes incredibly important. With alimited amount of talented human capital, it is difficult to pick out a profitable customer to cooperate with. The case studies the actual operation of the server division of the company every month for the last five years, and uses it as the basis for further customer screening. In this study, we use the profitability of each customer as dependent variable, and manpower input and net profit contribution that significantly affect profits as independent variables. Using customer characteristics as dummy variables, we construct a model with multiple regression analysis identifying different customers’profitability, and compared it with actual operating conditions. Next, we used the regression analysis model we obtained on forecastingfuture profitability; finally, we explored how many existing customers and new customers can meet the screening criteria. The study found that selecting different amount of customers’ firms will result in a different regression model, but if there are a closeamount of major clients, its results of the expected profit will be very similar in reality in terms of direction. However, deviation in terms of forecasting profits can be more reasonably larger. Finally, the study can be summarized as follows: (1) the regression model can provide direction about profit, but can not precisely predict the value of exceeding profit. (2) after a period of time, it is required to re-examine the regression model, adding the follow-up information on new business performance, to ensure its applicability of evaluating current operation situation. Keywords： Customer selection Profit estimation Regression analysis

在試題難易度分析上，傳統方式常以答對率或得分率的高低來認定試題的難易度。但答對率或得分率高低並不能真正表達應試者在答題時的難易感受程度。過去研究指出影響數學科試題難易度的主要因素有三：評量的數學內容、解題時的思考策略及解題所需的步驟數。 本論文針對影響數學科試題難易度的三個主要因素進行分析統計。以模糊統計的角度，提出非固定權重因子之二維模糊數。應用模糊數絕對距離的概念以摒除族群中異端值，最後針對不同族群進行難易指標分析與檢定。 / On the basis of difficulty analysis, the hit rate or the scoring rate are considered as the index of the difficulty of the questions traditionally. However, these rates can not represent how test takers feel when taking tests. According to the previous papers on this subject, they note three factors to affect the difficulty analysis on math questions the content of the evaluation, the strategies of question solving, and the required steps to solve a question. This research is focused on the three major factors which influence the difficulty of math examinations. With the angle of fuzzy statistics, a two-dimension fuzzy number will be presented with non-fixed weighted factors. By applying the absolute distance concept of fuzzy number, the extreme values are excluded. Consequently, the indices of difficulty from different groups can be tested and analyzed.