Basket-Sensitive Random Walk & Factorization Machine Recommendations For Grocery Shopping

While recommendation systems have been a hot topic for a long time now due to its success in business applications, it is still facing substantial challenges. As grocery shopping is most often considered as a real drudgery, many online stores provide a shopping recommendation system for their customers to facilitate this purchase process. However, there is still a large majority of people who still hesitate from doing their groceries online even though this form of shopping provides consumers with distinct advantages. Therefore, the goal of this paper is to investigate whether traditional collaborative filtering techniques are applicable in the domain of grocery shopping, and further improve its recommendations using extensive models and machine learning techniques. Hence, various CF-based models have been constructed including your traditional similarity-based collaborative filtering models, a basket-sensitive random walk model, and a basket-sensitive factorization machine. Here, we found that our basket-sensitive factorization machine comes out on top when it comes to recommending less popular items. However, due to its computational time, it remains to be a question whether this model is applicable in practical use.