Back to Details

analyze-data

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Pre-requisites

前置要求

  • Set up Holistics MCP
  • 完成Holistics MCP的搭建

Syntactic and descriptive

语法与说明层面

Always use 
explain_result
tool to syntactically/descriptively explain AQLs and their results. It can understand AQL accurately and provide accurate answers.
始终使用
explain_result
工具从语法/说明层面解释AQL及其结果。它可以准确理解AQL并提供精准的答案。

Diagnostic and predictive

诊断与预测层面

Run additional queries to analyze and provide diagnostic and predictive answers. For example, break down further, break down by other dimensions, compare different periods, etc. to answer questions such as "why did this number drop", "what was the main contributing factor", etc.
运行额外的查询来分析并给出诊断和预测类答案。例如，进一步拆解、按其他维度拆分、对比不同周期等，以此回答“这个数值为什么下降了”、“主要影响因素是什么”这类问题。

General guidelines

通用准则

  • Start with using 
    explain_result
    to understand the query, then proceed to further diagnostic/predictive analysis.
  • Beware of incomplete time periods when doing period comparisons.
    • For example, if today is 2025 Oct, and you are comparing the sales of 2025 Q4 to 2024 Q4, the sales of 2025 Q4 might be lower simply because it's only the beginning of the quarter. Check the SQL (included in output of 
      execute_aql
      , 
      execute_viz
      ) to verify.
  • 首先使用
    explain_result
    理解查询，再进行后续的诊断/预测分析。
  • 进行周期对比时注意时间段不完整的问题。
    • 例如，如果今天是2025年10月，你在对比2025年第四季度和2024年第四季度的销售额，2025年Q4的销售额较低可能仅仅是因为季度才刚开始。可查看
      execute_aql
      、
      execute_viz
      输出中包含的SQL来核实。