Product recommendation algorithms in the age of omnichannel retailing - An intuitive clustering approach

In today's omnichannel retailing world, product recommendations have become important in retailer strategy. Using big data to recommend complementary products can help improve customer service and thereby increase profitability. A common implementation for studying buying behaviour of customers uses a 0-1 matrix linking the customers to the products they have purchased in the past. However, this raw matrix does not automatically reveal buying patterns. Further processing of this matrix is necessary to find valuable information. In this work, we adopt an intuitive co-clustering algorithm for locating useful patterns in the matrix. The advantage of duplication of products in the clustering process will be shown. A further advantage of the algorithm from a managerial perspective is that it is intuitive rather than a black box type and thus may increase the chances of it being actually adopted.