尚未登入 請由此登入
 
 
 










知識分享平台Eshare檢視資訊View
返回前一頁Back
      [檢舉]

標題Title: Product form feature selection for numerical definition-based design
作者Authors: 楊智傑,陳鴻源..等
上傳單位Department: 多媒體與電腦娛樂科學系
上傳時間Date: 2010-7-28
上傳者Author: 楊智傑
審核單位Department: 多媒體與電腦娛樂科學系
審核老師Teacher: 楊智傑
檔案類型Categories: 論文Thesis
關鍵詞Keyword: Feature selection, support vector regression, support vector machine recursive feature elimination (SVM-RFE), Multiple linear regression, Kansei engineering, Product form design
摘要Abstract: For developing appealing products, it is important for designers to pin point critical product form features (PFFs) that influence consumers’ affective responses (CARs) toward a product design. Manual inspection of critical form features based on expert opinions has not proved to meet with the acceptance of consumers. In this paper, a product form feature selection model based on CARs and a numerical definition-based systematic approach (NSDA) is proposed. First, NSDA was used to generate an explicit numerical definition of product form design. Next, CARs were described using single adjectives as affective dimensions. The evaluation data of consumers can be gathered by a semantic differential (SD) experiment. The prediction models of CARs were constructed using support vector regression (SVR) and multiple linear regression (MLR). Two feature selection methods, namely SVR with support vector machine recursive feature elimination (SVM-RFE) and MLR with the stepwise procedure, were used to select critical form features. A case study of knife design is given to demonstrate the experimental results of the two feature selection methods. The results of our experiment show that the feature ranking obtained from SVM-RFE is very helpful to determine the importance of the form features. It is also possible to select a subset of features with a given predictive performance of the SVR model. On the other hand, the results of MLR with the stepwise procedure provide several useful statistics to analyze the form features. The effects of the form features for inducing specific CAR of knife design can also be interpreted by the standardized regression coefficients. Since only a small data set of knife design was adopted in this study, a further study using different kinds of product samples is worthy of ongoing investigation.

檔案名稱
FileName
檔案大小
Size
檔案格式
Format
瀏覽次數
Browses
下載次數
Downloads
2010_7_4ada4649.doc 1258Kb doc 463 238
文件中檔案:
 

開啟檔案Download
 
 
返回前一頁