Predicting Size and Fit in Fashion E-Commerce

Fashion e-commerce presents new challenges for retailers in predicting the size and fit of clothing items for their customers. Accurately predicting the size and fit of clothing items is essential for customer satisfaction and can reduce the likelihood of returns and exchanges. In literature, diverse approaches exist, varying from traditional methods (e.g., body scanning and 3D modeling) to more advanced techniques based on computer vision and/or deep learning. Machine learning-based techniques can potentially improve the overall accuracy of size and fit predictions in fashion e-commerce. This work is based on an existing framework known as K-Latent Factors with Metric Learning (K-LF-ML) but uses support vector machines (SVMs) for the final classification purpose. SVM can learn from customer preferences and past purchases to classify clothing items based on their size and fit, and it can help retailers to make more informed recommendations to their customers. We have performed several experiments to assess the performance of the proposed method using a publicly available dataset, RentTheRunWay. The empirical results demonstrate a better value for the metric Area Under the Curve (AUC) against the baseline method.