Paper: arXiv 2406.19399

Authors: Andrew Estornell, Stylianos Loukas Vasileiou, William Yeoh, Daniel Borrajo, Rui Silva

Abstract

In today’s competitive financial landscape, understanding and anticipating customer goals is crucial for institutions to deliver a personalized and optimized user experience. This has given rise to the problem of accurately predicting customer goals and actions. Focusing on that problem, we use historical customer traces generated by a realistic simulator and present two simple models for predicting customer goals and future actions – an LSTM model and an LSTM model enhanced with state-space graph embeddings. Our results demonstrate the effectiveness of these models when it comes to predicting customer goals and actions.

Complexity vs Empirical Score

  • Math Complexity: 2.5/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Street Traders — empirical and applied, lower math complexity

Why this score: The paper uses a straightforward LSTM model with handcrafted features on a simulated dataset, showing low mathematical complexity; however, it employs a realistic simulator and presents specific accuracy metrics for goal prediction, demonstrating solid empirical implementation.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Predict Customer Goals in Financial Services"] --> B["Methodology<br>LSTM & LSTM with Graph Embeddings"]
  B --> C["Data Source<br>Historical Traces from Simulator"]
  C --> D["Computational Process<br>Sequential Modeling & Prediction"]
  D --> E["Key Outcome<br>Accurate Prediction of Goals & Actions"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes two models for predicting customer goals and actions: a standard LSTM and an LSTM enhanced with state-space graph embeddings.

Source sentence
we use historical customer traces generated by a realistic simulator and present two simple models for predicting customer goals and future actions -- an LSTM model and an LSTM model enhanced with state-space graph embeddings.

Assumes 3

  • The synthetic simulator generates data that accurately reflects real banking customer behavior. data
    Source sentence
    By leveraging this simulator, they are able to create datasets that accurately represent the intricacies and dynamics of customer traces in a banking environment
  • A frequency threshold of 10 occurrences is sufficient to define meaningful nodes and edges in the state graph. model
    Source sentence
    The graph was constructed by examining all customer trajectories in the training dataset and defining nodes and edges as any state or transition action which appeared in more than 10 times. We found 10 to be the best frequency threshold when defining the graph
  • Customer sequences are represented using bag‑of‑words and one‑hot encodings. model
    Source sentence
    Specifically, we used a bag-of-words and one-hot encoding to represent the events as well as constructed a state graph to use as a GNN embedding in the LSTM model.

Data

  • semi‑synthetic dataset of customer interactions synthetic unknown
    Source sentence
    Our experimental setup consisted of the semi-synthetic dataset with 12 thousand customer interactions, split into 70% training, 15% validation, and 15% test sets.

Costs

Not applicable The paper reports no trading or execution result.

Finds 3

  • Goal prediction accuracy reaches 71% for the LSTM and 77% for the GNN+LSTM model. predictive-accuracy
    Source sentence
    ... Table 3: Accuracy for each approach when predicting agent type.
  • Customer type prediction accuracy is roughly 90% or higher for both models across income, fail behavior, and digital behavior categories. predictive-accuracy
    Source sentence
    LSTM 63% 89% 97% GNN+LSTM 70% 90% 97% Table 4: Accuracy for each approach when predicting customer type.
  • Trajectory prediction accuracy declines with horizon length, but GNN+LSTM consistently outperforms LSTM (e.g., 67% vs 52% for next 1 event). predictive-accuracy
    Source sentence
    LSTM 52% 40% 33% GNN+LSTM 67% 62% 49% Table 5: Accuracy for approach when predicting the next 1, 5, and 15 customer events.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 34 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive (1 by independent authors)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (26)

  1. Domain-independent generation and classification of behavior traces (2020) — link
  2. Predicting the next app that you are going to use (2015)
  3. Goal recognition design (2014)
  4. A Survey on Goal Recognition as Planning (2021)
  5. Probabilistic plan recognition using off-the-shelf classical planners (2010)
  6. Heuristic Online Goal Recognition in Continuous Domains (2017)
  7. Goal recognition design with stochastic agent action outcomes (2016)
  8. Stochastic Goal Recognition Design Problems with Suboptimal Agents (2022)
and 18 more
  1. New Metrics and Algorithms for Stochastic Goal Recognition Design Problems (2017)
  2. Accounting for Partial Observability in Stochastic Goal Recognition Design: Messing with the Marauder’s Map (2020)
  3. Goal Recognition Design in Deterministic Environments (2019)
  4. Long short-term memory (1997)
  5. Machine learning based customer churn prediction in banking (2020)
  6. A Prediction Model for Recognition of Bad Credit Customers in Saman Bank Using Neural Networks (2011)
  7. Predicting customer churn in banking industry using neural networks (2016)
  8. Bankruptcy Prediction Using Machine Learning Techniques (2022)
  9. Goal recognition via model-based and model-free techniques (2020)
  10. Investigating the Combination of Planning-Based and Data-Driven Methods for Goal Recognition (2023) — link
  11. What’s the Context? Implicit and Explicit Assumptions in Model-Based Goal Recognition. (2021)
  12. Action understanding as inverse planning (2009)
  13. Plan recognition in continuous domains (2018)
  14. Active goal recognition (2020)
  15. Epistemic plan recognition (2020)
  16. Plan Recognition as Planning Revisited. (2016)
  17. Adam: A method for stochastic optimization (2014) — link
  18. Distributional structure (1954)

0 of 26 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.