Modeling retail browsing sessions and wearables data

The advent of wearable non-invasive sensors for the consumer market has made it cost-effective to conduct studies that integrate physiological measures such as heart rate into data analysis research. In this paper we investigate the predictive value of heart rate measurements from a commercial wrist wearable device in the context of e-commerce. We look into a dataset comprised of browser-logs and wearables data from 28 individuals in a field experiment over a period of ten days. We are particularly interested in finding predictors for starting a retail session, such as the heart rate at the beginning of a web browsing session. We describe preprocessing tasks applied to the dataset and logistic regression and survival analysis models to retrieve the probability of starting a retail browsing session. Preliminary results show that heart rate has a significant predictive value on starting a retail session if we consider increased and decreased heart rate individual values and the time of day.