In the context of clothing and wearable products, fashion is prone to volatile trends. In clothing fashion there are different seasons of clothing, which account for some of the changing patterns. There are also trends on the scale of the general public, on both shorter time scales and longer time scales. The volatility of these trends poses an issue to conventional natural language processing techniques as well as machine learning approaches. Due to the frequent and unpredictable changes that can occur in a fashion context, models that cannot adapt eventually fail. Like our prior work [1], the model developed here predicts the category and subcategory of fashion items based on the textual contents of the title. The model developed is also capable of adapting to future changes in fashion including the addition of new terminology and changing popularity and classification for existing items. This paper covers some of the problems conventional natural language processing approaches face when tasked with classifying titles in a fashion context. It then covers a few potential approaches to dealing with the implementation of machine learning approaches for classification purposes, and why they fail in the given situation. Finally, this paper presents a solution in the form of a model utilizing feature hashing and the Passive-Aggressive classifier. The results show this model performs as well as the prior model, with a much better training time. This model also possesses the ability to adapt to future changes in fashion.