Research and implementation of a trend prediction model based on trend similarity for the changing trends of fashion elements in clothing

Abstract

Trend forecasting of clothing fashion elements is an important guide for product development and sales of garment companies. Existing work can only capture simple changing trend laws and patterns of mutual influence between trends but cannot give effective and practical guidance on the trend changes of clothing fashion elements. This paper uses user information to group rich fashion elements in a more accurate and meaningful way to predict the trend of future trends in fashion elements. By comparing the similarity between the recent trend changes and the historical trend information, we continuously evaluate the next change trend information from the similar historical trend information, learn the laws and patterns of clothing fashion element change trends and predict the future trend change direction. Our experiments show that the model proposed in this paper can effectively capture the changing laws of clothing fashion elements and the patterns that affect each other to predict the changing trends. Compared with the baseline method, the model has the best performance in MAE and MAPE indicators.