Uncovering Characteristic Response Paths of a Population

Abstract

We propose an approach to uncover characteristic response paths of a population from an individual-level multivariate time series dataset. The approach is based on a model that accommodates arbitrary and mixed set of distributions for the endogenous variables, accommodates intersession intervals and variables, and reliably estimates individuals’ parameters through statistical pooling by uncovering clusters of similar individuals. We show that using such a model one can distribute response of a target variable to an exogenous impulse over all possible activity sequences leading up to it. When a few such sequences explain majority of the response, they describe the population’s characteristic response paths to the impulse.

We apply the proposed approach to a customer touchpoint dataset from a large multi-channel specialty retailer and uncover six segments with different characteristic shopping paths leading to purchase in response to a marketing communication. These uncovered paths provide insights into the behavior of and optimal over-time communication strategy for customers in different segments.

Keywords: Response Paths, Segmentation, Multivariate Time-series Model, Mixture Model, Dynamic Programming.