Fashion Style Forecasts based on Different Price Ranges

Fashion trend forecasting is of great significance to both fashion workers and consumers. Although the existing methods can effectively capture the trend of fashion style, most of them are aimed at the change of clothing style trend, and the factors affecting the trend of style are not fully considered. To solve these problems, we introduce a method to capture fashion elements and discover styles in the products, and an end-to-end LSTM encoding and decoding framework that can accurately forecast future trends of fashion styles in different price ranges is proposed. We demonstrate our model in the Amazon data set and the results indicate that our model has the best forecasting performance compared with baselines menthods.