# Analysis canvases Organize analysis into steps, inspect intermediate results, and save and reuse your workflow. 使用指南 · 模块: product · 使用者 · 部署形态：资料未说明 · 已确认产品版本：资料未说明 · 来源：现有文档 Conversations work well for questions and answers. Use an analysis canvas when a question needs several steps, intermediate results, or repeated adjustments to assumptions. Each step is a node; data dependencies automatically connect nodes, and the whole process stays on the canvas. | Your situation | What to use | | ----------------------------------------------------------------------------------- | ---------------------------------------------------------- | | One question, one answer, perhaps a few follow-ups | [Conversations](https://docs.asktable.com/en/conversation) | | Inspect intermediate results, compare assumptions, and retain the reasoning process | Analysis canvases | | Monitor conclusions over time with regular refreshes | [Data dashboards](https://docs.asktable.com/en/dashboard) | Canvases and conversations share the same data semantics and permissions, so they use consistent business definitions. ## Before you start | Item | Requirement | | ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Permissions | Access to the required data sources. Analysis canvases require the appropriate license tier; the entry point is unavailable when not enabled. See [Commercial licensing](https://docs.asktable.com/en/admin/license) | | Prerequisite | A connected data source; local files to analyze must be no larger than 10 MB | | Entry point | Main navigation on the left → Analysis canvases | ## Create a canvas Click “Analysis canvases” in the main navigation on the left to open the “All canvases” list. Click “Create canvas” or open an existing one. Search canvases by name, or use “Sort” to order by last modified, name, or creation time. With no canvases, the page is empty; use the button above to create the first one. ## Work on the canvas **Start in the welcome area** A new canvas has a welcome area in the center with guidance and three example entry points: “Query sales data in a database”, “Import and analyze an Excel file”, and “Generate data visualizations”. Describe what you want to do in the input box; “Add attachment” is available on the left. You can also drag `.xlsx`, `.xls`, or `.csv` files directly onto the canvas to create Excel nodes after import. Dragging files works on both empty canvases and canvases with existing nodes. **Add nodes from the context menu** Right-click a blank area to open a menu with “Query data”, “Web search”, “Import Excel”, “Auto layout”, and “Delete”. Nodes can also be created directly from the input box. There are five node types: data, chart, Python, Excel, and web search. The input prompt changes with the node type: data nodes ask you to describe a query, chart nodes ask what chart you want, Python nodes ask about data processing, and web search nodes ask what data to search for. **Watch node execution** Nodes show their status directly, such as processing, searching the web, or generating a chart. Query nodes show their results with row and column counts when execution finishes. Expand a node to inspect its tools, such as reading table structures, viewing data sources, searching metadata, finding tables, exploring databases, executing SQL, updating data, generating charts, executing Python, and submitting code. **Inspect intermediate results in the detail panel** Select a node and open the “Detail panel”. Available tabs depend on node type: “Data”, “Chart”, “Code”, “Configuration”, “SQL”, “Logs”, and “Description”. The panel is empty when no node is selected. **Continue working from a node** Nodes provide actions to “Open details”, “Create chart” (abbreviated on the canvas), “Ask a follow-up”, “Refresh data”, and “Delete”. After selecting data, the canvas toolbar offers “Ask about selected data” and “Process selected data with Python”. **Adjust charts and connections** “Chart theme” provides a “Style” selector (“Modern” means rounded corners and smooth curves; “Tech” means square corners and sharp lines), a color palette selector including “Custom colors”, and a chart preference text box. For example, enter “Always show axis labels and units”. Connections are generated automatically from data dependencies. Use the toolbar to “Hide reference connections” or “Show reference connections”. **Discuss the whole canvas in the conversation panel** The conversation panel on the right has “Node details” and “Conversation history” entry points and a “New conversation” action. Its empty state provides guidance for starting a conversation. To ask about the whole canvas, choose “New conversation” in the conversation list and ask about the canvas's data in the input box. ## Analyze a sales spreadsheet Suppose you want to know “Which stores had declining sales last month?”: 1. Drag an Excel or CSV file containing dates, stores, and sales onto the canvas. Confirm that the imported node has the correct columns and data. 2. Select the data, click “Ask about selected data”, and enter “Compare sales by store for last month and the month before, and list the stores with the largest declines”. 3. Check the scope and intermediate results in the result node's detail panel, then click “Create chart” to visualize the changes by store. 4. To investigate further, ask “Break down these stores' declines by product category”. The original data, comparison results, and chart remain on the canvas so you can review the evidence for each step. ## Refresh data and reuse canvases The canvas menu at the top left offers “Refresh data” and “Clone canvas”. Refreshing fetches data nodes again and reports success or failure afterward; it does not automatically rerun every analysis step. Cloning creates and opens a copy, useful for testing another assumption while preserving the original analysis. ## Key constraints | Item | Value | Where to change it | | ---------------------- | ---------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------- | | Per-file import limit | 10 MB by default; larger files are rejected | File limits in the administration console | | Supported file formats | `.xlsx`, `.xls`, `.csv` | — | | Merged cells | Multirow compound headers are unsupported; imports with merged cells are rejected and must use the data source upload page instead | Upload through the data source page | | Duplicate column names | Renamed automatically, with renamed columns listed | — | | Web search | Requires prior configuration by a system administrator; otherwise unavailable | System settings | | Legacy chart nodes | Cannot be edited; delete and recreate them, or click “Upgrade” to create a new-format node alongside | — | ## Troubleshooting | Symptom | How to identify it | What to do | | ------------------------------- | ----------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- | | Node execution fails | A red error marker appears; expand it for details | Open details and check “Logs”; a common cause is SQL referencing a hidden field | | Chart rendering fails | The node reports failure and a reason | Generate the chart again or adjust it in “Configuration” | | Imported Excel data is empty | Row count is 0 or a message says the table is empty | Check whether the file contains only headers; for multiple sheets, confirm the selection in “Select sheets to import” | | Column-name error during import | Import is rejected for duplicate names or no valid header columns | Fix duplicate names in the source file first | | Data refresh fails | Refresh reports an error and node data is unchanged | Check that the source is still reachable; after source schema changes, first use “Resync” on the data source page | | Canvas does not open | A message says it does not exist or you lack access | Check that the account has project access; analysis canvases require the appropriate license tier | ## Next steps - Need a few metrics without the full process: [Conversations](https://docs.asktable.com/en/conversation) are faster - Monitor results over time: add them to a [data dashboard](https://docs.asktable.com/en/dashboard) - Let the team reuse this analysis: turn the method into a [skill](https://docs.asktable.com/en/agent/skill)