Methodology for fashion colour trend prediction based on grey theory

論文翻譯標題: 應用灰色理論於流行色彩趨勢預測之研究
  • 李 筑軒

學生論文: Doctoral Thesis


For design and manufacturing industries being able to grasp the fashion trend is an essential factor that leads to winning a sale However colour predicting process in many institutions are not visible to the public In order to provide colour trend to industries in advance a predicting method is proposed in this study In the method firstly a new Fuzzy C-Means clustering algorithm was used to separate the colour data being collected In this step two models of colour arrangement were separately applied in this study One is the least mean-square error (LMSE) model which was used to place the similar colour clusters within different time point together At the same time the other model of grey relational analysis will be used to find the closest colour combination Finally the grey model was adopted for predicting these two models In order to verify the probability of the system four data announced by Pantone from spring 2014 to fall 2015 were taken as the predicted samples and the colour for spring 2016 was predicted to compare with that in Pantone spring 2016 The initial results show that the system has a high accuracy for predicting colour with the LMSE model The residual modified model constructed with the colour samples rearranged with LMSE has the best-predicted result that ranged from 83 3%- 99 4% It indicates that the result obtained with the rearranged samples is higher than that without rearrangement Besides the accuracy of the grey predicted results with the residual modification would be more precise than the formal grey prediction model Moreover the value of mean squared error is quite low which was ranged from 0 000025-0 0277 On the other hand the predicted result with GRA model was ranged from 2 28%-98 9% indicating that the accuracy of the prediction result with GRA model still needs slight improvement Besides it is easy to find that the MSE value was ranged from 0 0001-0 5963 This result means that this system can generate a stable accuracy outcome in fashion colour prediction Therefore the current intelligent predicting system satisfies the criteria of capturing colour in trend for enterprises Moreover it enables industries to make decisions for selecting the colour trend
獎項日期2017 六月 26
監督員Shih-Wen Hsiao (Supervisor)