QiyanAgent 企小喵实证研究智能体 Download Client
Data Analysis Agent

Clarify the research question before asking the Data Analysis Agent to choose a method.

The Data Analysis Agent covers research design, variable preparation, method selection, model execution, robustness checks, and result interpretation. It supports econometrics, management analysis, and machine learning, and separates correlation, causality, descriptive findings, and model limitations.

Applies to QiyanAgent V3.5. Last updated: September 14, 2026.

Data Analysis Agent steps: choose a research design or analysis method, confirm built-in Skills, enter variable, sample, and model requirements, and send the task.

This is a QiyanAgent V3.5 interface screenshot. The callouts mark research-design and method selection, Skills configuration, task entry, and task submission.

Before You Start

  • State the research question, dependent variable, independent variables, control variables, and sample scope.
  • Confirm the data has been cleaned and can be read through a database or project file.
  • State whether you need OLS, Logit, fixed effects, DID, event study, RDD, IV/2SLS, PSM, or machine-learning methods.

Steps

  1. In the Data Analysis workspace, review packages for causal identification, paper replication, case learning, Stata, OLS, Logit, fixed effects, DID, event studies, RDD, IV/2SLS, PSM, mediation and moderation, SEM, regularized regression, random forests, gradient-boosted trees, SVM, K-means, PCA, model evaluation, and causal forests.
  2. You can choose an analysis method manually, or describe the research question, variables, and sample so the Agent can choose a suitable method. Configure the Stata environment only when it is needed.
  3. To review capabilities, open Settings → Skills and switch to the Data Analysis Agent. Built-in Skills sync automatically; use Import only to add method guidance.
  4. Enter an analysis prompt with data table, variable definitions, sample restrictions, model specification, output format, and robustness requirements.
  5. Click the send button and let the Data Analysis Agent draft the analysis plan, run models, and interpret results and risks.

Acceptance Checks

The output should include variable checks, model specification, result tables, interpretation, robustness suggestions, and risk notes. Interpretation should separate correlation, causality, descriptive findings, and model limitations.