AI Paper Atlas

Research map / Large language models / Agents, code and math

Why Do Open-Source LLMs Struggle with Data Analysis? A Systematic Empirical Study

Yuqi Zhu, Yi Zhong, Jintian Zhang, Ziheng Zhang, Shuofei Qiao, Yujie Luo, Lun Du, Da Zheng, Ningyu Zhang, Huajun Chen

AAAI 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 7 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
provide table information (columns, types, samples) in the prompt (ablation)Qwen2.5-7B prompting · QRDataaccuracy (%): 6.57 -> 7.54Improved
provide table information in the prompt (ablation)Qwen2.5-14B prompting · DiscoveryBenchaccuracy (%): 0.42 -> 0.00Hurt
add irrelevant extra data files to the input (ablation)Qwen2.5-7B prompting · QRDataaccuracy (%): 37.96 -> 34.55Hurt
add irrelevant extra data files to the input (ablation)Qwen2.5-14B prompting · QRDataaccuracy (%): 52.55 -> 52.07No clear effect
use code-specialized model (Qwen2.5-7B-Coder) instead of general instruct (ablation)Qwen2.5-7B · QRDataaccuracy (%): 39.71 -> 36.50No clear effect
use R1-distilled 7B model instead of instruct (ablation)Qwen2.5-7B · QRDataaccuracy (%): 39.71 -> 30.41Hurt
train on medium-length interaction turns rather than mixed turn lengths (ablation)Qwen2.5-7B SFT · QRData and DiscoveryBenchaccuracyImproved
train on all 5613 samples instead of the medium-turn subset (ablation)Qwen2.5-7B SFT · QRDataaccuracy (%): 49.15 -> 48.66Hurt
replace original reasoning with full R1 think traces (ablation)Qwen2.5-7B SFT · DiscoveryBenchaccuracyHurt
replace original reasoning with summarized reasoning (ablation)Qwen2.5-7B SFT · QRData and DiscoveryBenchaccuracyImproved
increase per-turn token budget (ablation)Qwen2.5-7B SFT · DiscoveryBenchaccuracyMixed
train on medium plus hard difficulty data (ablation)Qwen2.5-7B SFT · DiscoveryBenchaccuracy (%): 18.83 -> 23.01Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy