Ask-the-data probe report

dc.creatorPalacios, P.
dc.creatorJuarez, H.
dc.creatorMwanzia, L.
dc.date2025-12
dc.date2026-01-30T19:14:39Z
dc.date2026-01-30T19:14:39Z
dc.date.accessioned2026-06-27T17:59:18Z
dc.descriptionCIP Dataverse serves as the primary repository for CIP experimental datasets, but functions as a noninteractive, download-oriented platform lacking built-in analysis and visualization capabilities. This creates significant barriers for researchers who need coding expertise to perform data analysis, limiting accessibility and reproducibility of findings for domain experts who are not data scientists. Ask-the-Data (ATD) is a Shiny for Python application that addresses these challenges by integrating interactive data exploration with AI assistance. The tool connects to datasets via DOI links from CIP Dataverse, enables users to select tables and engage in natural language conversations with a Pandas DataFrame agent, and generates visualizations and statistical analyses without requiring coding expertise. Key results include: (1) successful deployment of a web-accessible application at https://askthedata.cipotato.org/; (2) integration of LangChain-powered conversational AI with pandas, scikit-learn, and statsmodels libraries; (3) capability to generate comprehensive markdown reports including visualizations, data tables, and conversation transcripts. The tool democratizes data science by enabling researchers to perform advanced analyses through natural language queries.
dc.formatapplication/pdf
dc.identifierhttps://hdl.handle.net/10568/180983
dc.identifier.urihttp://hdl.handle.net/123456789/151619
dc.languageen
dc.rightsOpen Access
dc.sourcePalacios, P.; Juarez, H.; Mwanzia, L. 2025. Ask-the-data probe report. CGIAR Accelerator on Digital Transformation; International Potato Center.
dc.subjectdigital agriculture
dc.subjectopen data
dc.subjectartificial intelligence
dc.subjectdata analysis
dc.subjectlarge language
dc.subjectmodels
dc.titleAsk-the-data probe report
dc.typeReport

Archivos