Search

Tag: #data-analytics39 results

AIDABench: AI Data Analytics Benchmark

AIDABench: AI Data Analytics Benchmark

AIDABench introduces 600+ diverse tasks for end-to-end AI evaluation in data analytics, covering QA, visualization, and file generation across real-world heterogeneous data. Even top models like Claude Sonnet 4.5 score only 59.43% pass@1, revealing significant challenges. The benchmark is publicly available on GitHub.

ArXiv AIResearchMar 18#benchmark#data-analytics
Page 1 of 4