Short Term Forecasting of Western Clothing Using Machine Learning Tools

Abstract

Fashion trend forecasting in its essence started being conducted through human-based processes which depended quite majorly on the designers' artistic point of views. Nevertheless, with the rapid development of data science and the growing obtainability of data inputs coming from consumers, the probability of the usage of machine learning equipment to forecast fashion trends is captivating increasing interests among young academia in the industry of fashion. The study assessed the use of computing tools in order to achieve results of considerable relevance. Datasets from e-commerce giants like Amazon were implemented in order to achieve a result which could be used for predictions. We can also see the differences and similarities of the evaluations of traditional human-based fashion related trend forecasts and the ones resulted by computational equipment. Based on the evaluation of the result made by the one generated using time series tools gave very similar results to what would be considered a human-based process. The discovery of this evaluation fulfils an important study gap related to the feasibility of the usage of machine learning tools for the purpose of fashion companies' creative activities.