基于岭回归方法的羊毛衫洗后特征感性评价

服装产品不同特征的感性评价之间常常存在相互制约的共线性关系,为建立它们之间的量化关系,以17件羊毛衫作为实验对象,通过现场问卷得到30名成年女性对羊毛衫整体印象及9种特征的感性评价数据,使用岭回归方法建立羊毛衫多种特征与整体印象感性评价之间的量化关系。结果显示,羊毛衫9种特征对整体印象感性评价的影响权重排序为:大身平整度、弹性、蓬松度、尺寸比例、领口形态、下摆形态、起毛起球、磨损程度、袖窿接缝平整度,它们之间的岭回归方程拟合精度为0.977。该结果不仅可以帮助产品研发企业精准定位消费者的关注重点,而且可预测消费者对产品功效的感性评价。研究证明了岭回归方法适用于建立具有共线性关系的服装感性评价的量化关系,可广泛应用于不同产品的感性工学研究。

Study on sensory evaluation of performance of washed wool sweaters based on ridge regression method

For clothing products, there is usually a collinearity relationship between different sensory properties evaluation. In order to obtain the relationship, 17 wool sweaters were used as the experimental objects, by using on-the-spot questionnaire survey, 9 specific sensory evaluations from 30 women assessors were obtained, which were then analyzed using the ridge regression method, so as to achieve the quantitative relationships among the overall impression and different sensory property evaluations of clothing. The weights of different sensory properties on the overall impression are sorted as flatness, elasticity, bulkiness, size proportion effect, neckline shape, hem shape, fuzz and pilling, damage, armhole joint shape. In addition, the accuracy of regression fitting between the predicted value from ridge regression equation and the actual surveyed value reaches 0.977. It does not only predict the consumers' sensory evaluation of product, but also help the enterprises to accurately focus on the consumers about products. In conclusion, ridge regression is proved to be a good method to study the quantitative relationship of sensory property evaluation with collinear relationship, and it can be widely used in kansei engineering of a variety of garments.