![[2602.21634] AgentLTV: An Agent-Based Unified Search-and-Evolution Framework for Automated Lifetime Value Prediction](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.21634] AgentLTV: An Agent-Based Unified Search-and-Evolution Framework for Automated Lifetime Value Prediction
Summary
AgentLTV introduces an agent-based framework for automated Lifetime Value (LTV) prediction, enhancing model discovery and performance in various decision scenarios.
Why It Matters
Lifetime Value prediction is crucial for businesses in advertising and e-commerce. The AgentLTV framework streamlines the modeling process, making it adaptable and efficient, which is essential for practitioners facing diverse data patterns and complex decision-making environments.
Key Takeaways
- AgentLTV automates LTV modeling using an agent-based approach.
- Monte Carlo Tree Search (MCTS) aids in adapting to new data patterns quickly.
- Evolutionary Algorithms (EA) enhance model stability and refinement.
- The framework has shown improved performance in ranking and error metrics.
- Deployment diagnostics ensure readiness and effectiveness in real-world applications.
Computer Science > Machine Learning arXiv:2602.21634 (cs) [Submitted on 25 Feb 2026] Title:AgentLTV: An Agent-Based Unified Search-and-Evolution Framework for Automated Lifetime Value Prediction Authors:Chaowei Wu, Huazhu Chen, Congde Yuan, Qirui Yang, Guoqing Song, Yue Gao, Li Luo, Frank Youhua Chen, Mengzhuo Guo View a PDF of the paper titled AgentLTV: An Agent-Based Unified Search-and-Evolution Framework for Automated Lifetime Value Prediction, by Chaowei Wu and 8 other authors View PDF HTML (experimental) Abstract:Lifetime Value (LTV) prediction is critical in advertising, recommender systems, and e-commerce. In practice, LTV data patterns vary across decision scenarios. As a result, practitioners often build complex, scenario-specific pipelines and iterate over feature processing, objective design, and tuning. This process is expensive and hard to transfer. We propose AgentLTV, an agent-based unified search-and-evolution framework for automated LTV modeling. AgentLTV treats each candidate solution as an {executable pipeline program}. LLM-driven agents generate code, run and repair pipelines, and analyze execution feedback. Two decision agents coordinate a two-stage search. The Monte Carlo Tree Search (MCTS) stage explores a broad space of modeling choices under a fixed budget, guided by the Polynomial Upper Confidence bounds for Trees criterion and a Pareto-aware multi-metric value function. The Evolutionary Algorithm (EA) stage refines the best MCTS program via islan...
