DS-STAR: Data Science Agent for Solving Diverse Tasks across Heterogeneous Formats and Open-Ended Queries
#DS-STAR #Data Science Agent #Large Language Models #AI #arXiv #Heterogeneous Data #Open-Ended Queries #Benchmark Performance
📌 Key Takeaways
- DS-STAR processes and integrates data across diverse, heterogeneous formats
- The agent generates comprehensive research reports for open-ended queries
- DS-STAR outperforms existing models in complex multi-file processing tasks
- Generated reports are preferred over best baseline models in 88% of cases
📖 Full Retelling
Researchers Jaehyun Nam, Jinsung Yoon, Jiefeng Chen, Raj Sinha, Jinwoo Shin, and Tomas Pfister introduced DS-STAR, a specialized data science agent, in a paper published on arXiv on September 26, 2025, with the latest revision released on February 24, 2026, to address the limitations of existing AI agents in handling complex real-world data science workflows. The DS-STAR agent represents a significant advancement in artificial intelligence by demonstrating the ability to process and integrate data across diverse, heterogeneous formats while generating comprehensive research reports for open-ended queries. Unlike previous approaches that often struggle with the complexity of real-world workflows requiring multiple data sources and synthesis of insights, DS-STAR bridges this critical gap in the field. The researchers conducted extensive evaluations showing that DS-STAR achieves state-of-the-art performance on four benchmarks: DABStep, DABStep-Research, KramaBench, and DA-Code. Particularly notable is its significant superiority over existing baseline models in difficult QA tasks requiring multi-file processing, and its generated data science reports were preferred over the best baseline model in over 88% of cases during testing.
🏷️ Themes
Artificial Intelligence, Data Science, Machine Learning
📚 Related People & Topics
Artificial intelligence
Intelligence of machines
# Artificial Intelligence (AI) **Artificial Intelligence (AI)** is a specialized field of computer science dedicated to the development and study of computational systems capable of performing tasks typically associated with human intelligence. These tasks include learning, reasoning, problem-solvi...
Large language model
Type of machine learning model
A large language model (LLM) is a language model trained with self-supervised machine learning on a vast amount of text, designed for natural language processing tasks, especially language generation. The largest and most capable LLMs are generative pre-trained transformers (GPTs) that provide the c...
Entity Intersection Graph
Connections for Artificial intelligence:
🏢
OpenAI
11 shared
🏢
Nvidia
7 shared
🏢
Anthropic
4 shared
👤
Wall Street
4 shared
🌐
Machine learning
3 shared
Original Source
--> Computer Science > Artificial Intelligence arXiv:2509.21825 [Submitted on 26 Sep 2025 ( v1 ), last revised 24 Feb 2026 (this version, v4)] Title: DS-STAR: Data Science Agent for Solving Diverse Tasks across Heterogeneous Formats and Open-Ended Queries Authors: Jaehyun Nam , Jinsung Yoon , Jiefeng Chen , Raj Sinha , Jinwoo Shin , Tomas Pfister View a PDF of the paper titled DS-STAR: Data Science Agent for Solving Diverse Tasks across Heterogeneous Formats and Open-Ended Queries, by Jaehyun Nam and 5 other authors View PDF HTML Abstract: While large language models have shown promise in automating data science, existing agents often struggle with the complexity of real-world workflows that require exploring multiple sources and synthesizing open-ended insights. In this paper, we introduce DS-STAR, a specialized agent to bridge this gap. Unlike prior approaches, DS-STAR is designed to (1) seamlessly process and integrate data across diverse, heterogeneous formats, and (2) move beyond simple QA to generate comprehensive research reports for open-ended queries. Extensive evaluation shows that DS-STAR achieves state-of-the-art performance on four benchmarks: DABStep, DABStep-Research, KramaBench, and DA-Code. Most notably, it significantly outperforms existing baseline models especially in hard-level QA tasks requiring multi-file processing, and generates high-quality data science reports that are preferred over the best baseline model in over 88% of cases. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2509.21825 [cs.AI] (or arXiv:2509.21825v4 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2509.21825 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Jaehyun Nam [ view email ] [v1] Fri, 26 Sep 2025 03:38:12 UTC (955 KB) [v2] Mon, 29 Sep 2025 03:14:15 UTC (955 KB) [v3] Thu, 2 Oct 2025 08:28:58 UTC (955 KB) [v4] Tue, 24 Feb 2026 08:22:41 UTC (2,482 KB) Full-text links: Access Paper: View a PDF of the paper titled DS-STAR: Dat...
Read full article at source