Generating In-store Customer Journeys from Scratch with GPT Architectures

TL;DR

This paper introduces a Transformer-based method to generate realistic in-store customer journeys and purchasing behaviors, improving accuracy over traditional models and reducing training data needs through fine-tuning.

Contribution

It presents a novel application of GPT-2 architecture for simulating retail customer trajectories and behaviors from scratch, with effective fine-tuning strategies.

Findings

Outperforms LSTM and SVM in trajectory and purchase behavior generation

Fine-tuning reduces training data requirements significantly

Achieves more accurate customer journey simulations

Abstract

We propose a method that can generate customer trajectories and purchasing behaviors in retail stores simultaneously using Transformer-based deep learning structure. Utilizing customer trajectory data, layout diagrams, and retail scanner data obtained from a retail store, we trained a GPT-2 architecture from scratch to generate indoor trajectories and purchase actions. Additionally, we explored the effectiveness of fine-tuning the pre-trained model with data from another store. Results demonstrate that our method reproduces in-store trajectories and purchase behaviors more accurately than LSTM and SVM models, with fine-tuning significantly reducing the required training data.