A model-based embedding technique for segmenting customers

We consider the problem of segmenting a large population of customers into nonoverlapping groups with similar preferences, using diverse preference observations such as purchases, ratings, clicks, and so forth, over subsets of items. We focus on the setting where the universe of items is large (ranging from thousands to millions) and unstructured (lacking well-defined attributes) and each customer provides observations for only a few items. These data characteristics limit the applicability of existing techniques in marketing and machine learning. To overcome these limitations, we propose a model-based embedding technique, which takes the customer observations and a probabilistic model class generating the observations as inputs, and outputs an embedding—a low-dimensional vector representation in Euclidean space—for each customer. We then cluster the embeddings to obtain the segments. Theoretically, we derive precise necessary and sufficient conditions that guarantee asymptotic recovery of the true segments under a standard latent class setup. Empirically, we demonstrate the speed and performance of our method in two real-world case studies: (a) up to 84% improvement in accuracy of new movie recommendations on the MovieLens data set and (b) up to 8% improvement in the performance of similar product recommendations algorithm on an offline data set at eBay. We show that our method outperforms standard latent class, empirical Bayesian, and demographic-based techniques.