EventGPT predicts banking customer actions with minimal data
EventGPT: Next-Event Pretraining on Banking Customer Histories
Machine Learning
Summary
Banks want to know what their customers will do next, like what kind of transaction they might make. The authors created EventGPT, a computer program that learns from past customer activity to guess what will happen next without needing special labels. This program works quickly, learns from a lot of information about transactions and locations, and can predict future actions quite accurately. EventGPT can also help improve other banking prediction tasks when used together with traditional data.
What this means in practice
- •For banking data scientists: Predict customers’ next financial actions with minimal labeled data by using pretrained EventGPT on transaction histories.
- •For financial product teams: Improve product-targeting models by combining EventGPT’s embeddings with traditional features for more accurate client behavior forecasts.
Tested on one dataset.
Authors
Denis Burakov
Abstract
A model pretrained to predict a customer's next event can estimate the probability of future events without training on their labels and provide a reusable representation for supervised downstream banking tasks. We pretrain EventGPT, a hierarchical causal transformer, by next-token maximum likelihood estimation on the public Multimodal Banking Dataset (MBD-mini: 100,224 clients, 38.8 million transactions and 66.3 million geolocation pings). Keeping one state per event, EventGPT trains in about two hours on one L4 GPU at windows up to 8,192 tokens. From its next-event distribution, with no labels, EventGPT scores five targets of the form "the client makes a transaction of a given kind for the first time next month" at ROC-AUC 0.79 to 0.92. Gradient boosting on 114 engineered features needs 10,000 to 300,000 labeled client-months to match it on four targets and stays below it on the fifth. For product prediction, frozen EventGPT embeddings score below the engineered features (mean ROC-AUC 0.787 against 0.808 over five test folds), and a rank-averaged ensemble of EventGPT, profile fields and token counts with the engineered features adds 0.014 to 0.017, of which 0.007 to 0.016 is EventGPT's own. The code and pretrained weights are released.