The presence of missing data in databases is common, which can make difficult or even prevent their analysis. Therefore, it is necessary to address them, for example, by filling in values or removing records or columns. This paper aims to present a strategy to facilitate the handling of missing data. To achieve this goal, the strategy uses an LLM (Large Language Models) for generating code and conducting data analysis, and 3D visualizations with Augmented Reality to understand the data missing. This strategy has been implemented in a case study, where users can visually explore, measure, and handle the data missing. The results indicate that the strategy can facilitate the analysis and handling of missing data; however, an increase in the complexity of visualizations and handling of data missing was observed as the scale of the datasets increased.

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

A Strategy Utilizing an LLM and Augmented Reality for Handling the Missing Data: A Case Study Using Unity, Vuforia and ChatGPT

  • Enio Vicente de Limas,
  • Rita de Fátima Rodrigues Guimarães,
  • Bianchi Serique Meiguins,
  • Leonardo Chaves Dutra da Rocha,
  • Diego Roberto Colombo Dias,
  • Marcelo de Paiva Guimarães

摘要

The presence of missing data in databases is common, which can make difficult or even prevent their analysis. Therefore, it is necessary to address them, for example, by filling in values or removing records or columns. This paper aims to present a strategy to facilitate the handling of missing data. To achieve this goal, the strategy uses an LLM (Large Language Models) for generating code and conducting data analysis, and 3D visualizations with Augmented Reality to understand the data missing. This strategy has been implemented in a case study, where users can visually explore, measure, and handle the data missing. The results indicate that the strategy can facilitate the analysis and handling of missing data; however, an increase in the complexity of visualizations and handling of data missing was observed as the scale of the datasets increased.