Data-Analysis-Agent

Data‑Analysis‑Agent is an open‑source, conversational data analysis assistant. You can use it online, download a desktop version, or run it locally from the command line. The tool accepts Excel, CSV, SQLite, MySQL, PostgreSQL, or SQL Server files and lets you ask questions in natural language.

When you upload or connect a data source, the Agent first previews the dataset, showing sheets, columns, and sample rows. This preview step helps reduce misunderstandings before the AI starts working. Next you pose a question in plain English, such as “What is the sales gap by region and product category?” The Agent automatically writes and executes the necessary SQL, then presents the result in a single conversation. If you want visual insight, you can request a chart. The system offers forty‑three smart chart types, including comparison, trend, distribution, geographic, relationship, and share visualisations, so you can quickly see patterns without leaving the chat.

The platform also supports advanced analytics. It can perform anomaly detection, percentile calculations, clustering, decision tree modeling, regression, and funnel analysis. For time‑series data, you can choose from ARIMA, SARIMA, Prophet, VAR, or GRU workflows. Every conclusion is paired with evidence: the SQL query, data quality checks, and metric definitions, allowing you to verify each step. If you need more than one analyst, the system lets you create multiple Agents, schedule tasks, and add manual approval nodes. The workflow engine assigns tasks based on dependencies and can pause or resume analyses automatically.

Reporting is integrated into the conversation. You can export the finished analysis as cleaned Excel, Word reports, or slide decks with built‑in styles. The Agent also provides a business canvas that includes template SWOT, BCG, and business model canvases; these can be edited in draw.io and exported. Hooks let you trigger actions, such as sending a message or exporting a file, at specific points in the dialogue. If you use Feishu, you can read and write multi‑dimensional tables directly in a group chat.

Key features

• Upload Excel, CSV, or connect to SQLite, MySQL, PostgreSQL, SQL Server
• Natural‑language query that auto‑generates and runs SQL
• 43 smart chart types covering comparison, trend, distribution, geography, relation, and share graphs
• Advanced statistical models: anomaly detection, clustering, decision trees, regression, funnel analysis
• Time‑series forecasting with ARIMA, SARIMA, Prophet, VAR, GRU