多準則決策方法具備了準則集合、方案集合等項目，可幫助決策者在決策問題中根據實際狀況需求，去評估最理想方案。軟體專案成本對於軟體開發決策者而言，包含多重評估的準則，不同決策者對於軟體開發的成本需求問題各有不同，其問題本身就具備了多準則以及方案評估的特質，因此本研究將軟體成本因子的評選歸類於多準則決策問題，採用公開軟體成本COCOMO資料庫[51]做為實驗案例。在進行軟體專案成本評估的過程中，常因為決策者主觀判斷性以及軟體成本準則之因素導致軟體專案成本評估更加困難，因此本研究主要提出（1）屬性篩選，資料集屬性間會互相干擾並影響最終之評估，本研究將資料集屬性做篩選，找出影響軟體成本顯著性高之屬性；（2）模糊TOPSIS模式，包含：（a）理想解類似度順序偏好法（TOPSIS）結合模糊理論找出高風險成本群方案；（b）導入資訊融合技術於權重計算，依照決策者使用情境將準則間的大小分配權重；（c）在實驗過程中準則對於每位決策者的平均偏好程度，並選擇最高偏好程度之方案為佳選擇。經實證結果顯示，本研究提出的模糊TOPSIS模式可以提供決策者在評估符合最理想的方案時，重要之決策參考。 Multi-criteria decision-making includes multi-criteria and program evaluations. This method can help decision-makers to make decision based on actual situations, providing decision-makers with an optimal program. Software project cost for decision-makers, which includes multi-estimate criteria and multi-program. Therefore, evaluating software cost is classified as a multi-criteria decision-making problem. This study uses the software project cost evaluation case for a study case. The contributions of this study are as follows, 1) Use the decision tree J48 to filter attributes. We can find the potential interaction between attributes, determine how attributes affect the final evaluation, and point out the important attribute. 2) The study proposes a Fuzzy Q-uantiles Method to fuzzfy attributes. It can make the results more reasonable. 3) Use the Fuzzy Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) model, including (a) using TOPSIS and Fuzzy theory to identify high-risk costs, (b) using information fusion technology into weight calculations. 4) Use a Constructive Cost Model database to verify. The database is a public database; it’s including 63projects and 16 attributes.Analytical results show that the fuzzy TOPSIS model can solve the above problem and help decision-makers evaluate software cost effectively.
