運用類神經網路於系統傢俱業裝修成本預測之研究

Using Neural Network for Decoration Cost Prediction of System Furniture Industry

楊春敏(Chun-Min Yang)

指導教授 ： 車振華

國立臺北科技大學/管理學院/工業工程與管理系EMBA班/碩士(2008年)

摘要

明確的預測室內裝修工程所需費用，一直是業主發包單位亟欲達成的目標，以創意為先專案製造為主的室內裝修產業，出現以客製化產品為主的連鎖門市經營-系統傢俱，系統傢俱以其規格化生產的構想，加上可拆卸、可組裝、可擴充性，達到裝潢施工快速，組合變化多樣的優點，漸為國人所喜愛。而製程上將垃圾及噪音留置工廠，使現場達到零污染，取代傳統木作裝潢方式，施工前也有專業的設計與規劃，如機能設計、空間規劃。其空間動線的舒暢感，整體裝潢的設計、比例、色彩搭配，是賞心悅目的關鍵。雖然連鎖業興起，業者開始以現代化的經營管理模式，但在實務上，顧客裝修需求在考量整體性前提下，仍需承包櫥櫃工程之外的工程項目，以利接單的競爭力。本研究希冀透過某系統傢俱業者之業務資料，針對其顧客之歷史消費記錄，進行資料分群，再應用資料探勘中具有學習能力的類神經網路(Artificial Neural Network)技術，參考過去期間顧客使用的產品別做為輸入源，推導出個案成本預測參考值，並訂定學習目標。本研究建立了經由學習所得的類神經網路模型，在接單前即能有效預測個案成本，幫助企業藉以準確找出可接單的客群以提昇整體績效。

關鍵字

系統傢俱 ； 類神經網路 ； 成本預測

並列摘要

An accuret cost estimation for interior decoration is always a specific target that the project owners desire to achive. In the interior decoration consturction industry, which now is focused on project base but principled with creation, we find a kind of chain-stores which sells customized produects founded; it is so called System Furniture. System Furniture is more and more popolar because its advantages: demountability, fabricatability and expandability cause to shorten the construction period as well as flexible in designing. Furthermore, when producing, no environment polution in construction area but keep garbage and noise in factory to replace traditional cabinetmaker’s work. Plus a professional designing and planning before construction, for example, functionable designing and space planning, does not only make people comfortable in the space but also harmonize creation, scale and color; these all are key factors caused the comfort and beauty. Although most of business owners manage the business in modern style due to the blooming of chain-stores, in practice, to consider entairly, the owners should contract to other constructions except cabinet construction to strength the competence to deal the case. The research expects to conclude a reference ratio for case cost and decide a learning target via applying the technical of Artificial Neural Network in past input of System Furniture owner’s product code for data analysis. The rescher establish a learned model of Neural Network to estimate Decoration Cost Prediction effectively before dealing a case. It hopes to assist business owner to figuer out these potential customers to upper its
whole business performance.

並列關鍵字

System Furniture ； Neural Network ； Cost Prediction

參考文獻

【22】 林照挺，以特徵式模型建構塑膠射出成形產品成本預估及產品雛型設計價值分析系統，碩士論文，台北科技大學工業工程與管理系，台北，2006。

【25】 楊金聲，利用類神經網路與線性迴歸進行成本預測之研究-以印刷電路板產業為例，碩士論文，中原大學資訊管理學系，桃園，2005。

【28】 Bode, J., “Decision support with neural networks in the management of research and development: concepts and application to cost estimation,” Information & Management, vol. 34, 1998, pp.33-40

【29】 Clark, P. and Niblett, T., “The CN2 Induction Algorithm,” Machine Learning, Vol. 3, 1989, pp.261-283.

【31】 Everitt, B. S., 1993, “Cluster Analysis”, John wiiley & Sons, Inc.