Scholarlink

<p>The growing integration of data-driven practices across disciplines has increased the need for accessible data analytics training, particularly for learners without formal programming backgrounds. In this context, this study investigates the educational potential of Low-Code and No-Code (LCNC) solutions powered by Large Language Models (LLMs) in enabling such learners to perform data analysis through natural language interaction. An empirical study was conducted across undergraduate and graduate courses at a four-year institution, where students with limited programming experience were assigned structured data analysis tasks using LLM-generated R code. Their engagement, performance, and interaction with the tools were systematically evaluated. The results show that most students were able to successfully complete the assigned tasks, demonstrating the effectiveness of LLM-powered approaches in lowering technical barriers. However, performance was strongly influenced by factors such as prompt clarity, the level of instructional support, and the availability of guided troubleshooting. The findings highlight both the strengths and limitations of LLM-powered LCNC tools in supporting learning, particularly in relation to reliability and user dependency on prompt design. Overall, LLM-based LCNC tools show significant promise in enhancing accessibility to data analytics education, but their effective integration requires emphasis on prompt engineering skills and critical evaluation of generated outputs. These insights provide practical guidance for incorporating AI-assisted tools into higher education and preparing students to navigate emerging LCNC environments effectively.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Large language model based low code platforms for data analytics in higher education

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月5日

查看全文
  • Nasir Saeed,
  • Farman Ullah

摘要

The growing integration of data-driven practices across disciplines has increased the need for accessible data analytics training, particularly for learners without formal programming backgrounds. In this context, this study investigates the educational potential of Low-Code and No-Code (LCNC) solutions powered by Large Language Models (LLMs) in enabling such learners to perform data analysis through natural language interaction. An empirical study was conducted across undergraduate and graduate courses at a four-year institution, where students with limited programming experience were assigned structured data analysis tasks using LLM-generated R code. Their engagement, performance, and interaction with the tools were systematically evaluated. The results show that most students were able to successfully complete the assigned tasks, demonstrating the effectiveness of LLM-powered approaches in lowering technical barriers. However, performance was strongly influenced by factors such as prompt clarity, the level of instructional support, and the availability of guided troubleshooting. The findings highlight both the strengths and limitations of LLM-powered LCNC tools in supporting learning, particularly in relation to reliability and user dependency on prompt design. Overall, LLM-based LCNC tools show significant promise in enhancing accessibility to data analytics education, but their effective integration requires emphasis on prompt engineering skills and critical evaluation of generated outputs. These insights provide practical guidance for incorporating AI-assisted tools into higher education and preparing students to navigate emerging LCNC environments effectively.