Predicting Next Shopping Stage for E‑Commerce Applications

E-commerce applications are almost ubiquitous in our day to day life. However, most of them have little to no adaptation to user needs, which can cause lower conversion rates or unsatisfied customers.

We propose a Machine Learning system which learns the user behaviour from previous sessions and predicts useful metrics for the next session. These metrics can be used to better target the customer, by offering better deals, personalized notifications, placing smart ads and so on.

The data used for the learning algorithm is extracted from Google Analytics Enhanced E-commerce. The learning model is a double recurrent neural network. It predicts for each session a probability score of the defined target classes: browse products, add a product to cart, checkout or complete transaction.