A machine learning model for predicting colour trends in the textile fashion industry in south-west Nigeria

Fashion is primarily based on adoption of trends by customers in the textile industry. Fashion trend forecasting is a complex process that aims at identifying future preferences of customers. The textile fashion industry is volatile, trends change very quickly. Fashion trend must be closely followed to increase sales amount. Colour forecasting is considered as one of the significant driving force in the textile fashion industry. If the supply of an item surpasses its demand, it would remain unsold thereby generating loss for the industry. How do we assist the textile manufacturing industries in solving the problem of under/over stocking? Tackling this question from a data driven vision perspective, we developed a model to forecast visual colour trends. In this study, a model for colour demand forecasting in the textile industry developed. This study involves the real life application of the model using real demand values for textile clothing by customers in the south west zone of Nigeria. We forecast future purchases based on historical demand data. Two approaches were combined for forecasting colour trends in the textile industry. The developed model first used the Convolutional Neural Network (CNN) for extracting hidden information/features/patterns from the image dataset. The extracted features were then applied to K- means algorithm for extracting the colours. The study proved that the two approaches used performed excellently well and that accurate colour forecasting can significantly enhance productivity and generate more sales for the textile industry