AskTable Canvas User Guide: Building Powerful Data Analysis Workflows

Canvas is AskTable's core feature, transforming data analysis from linear conversations into non-linear thinking modeling. This article provides a comprehensive guide to using Canvas.

---

1. What is Canvas?

1. Canvas Concept

Canvas is a visual data analysis workspace where you can:

• •Create multiple analysis nodes
• •Establish dependencies between nodes
• •Execute multiple analysis tasks in parallel
• •Organize and manage analysis workflows

2. Why Do We Need Canvas?

Limitations of Traditional Conversational Analysis:

• •❌ Linear flow, cannot parallelize
• •❌ Context easily lost
• •❌ Difficult to manage complex analysis
• •❌ Cannot reuse intermediate results

Canvas Advantages:

• •✅ Non-linear thinking modeling
• •✅ Visual dependency relationships
• •✅ Parallel task execution
• •✅ Reuse intermediate results
• •✅ Team collaboration

3. Canvas vs Conversation

---

2. Canvas Core Concepts

1. Node

A Node is the basic unit in Canvas, with each node executing a specific task.

Node Types:

• •Data Node: Data query node
• •Chart Node: Chart visualization node
• •Python Node: Python code execution node
• •Excel Node: Excel import/export node
• •Web Search Node: Web search node

2. Dependencies

Dependencies can be established between nodes:

• •Child nodes depend on parent node outputs
• •Child nodes can only execute after parent nodes complete
• •Supports multiple parent nodes

Example:

Data Node (Query Order Data)
  ├─> Chart Node (Sales Trend Chart)
  ├─> Chart Node (Regional Distribution Chart)
  └─> Python Node (Calculate Growth Rate)
        └─> Chart Node (Growth Rate Chart)

3. Data Flow

Data flows between nodes:

• •Data Node outputs DataFrame
• •Chart Node receives DataFrame, generates chart
• •Python Node receives DataFrame, executes code, outputs new DataFrame

---

3. Node Types in Detail

1. Data Node

Function: Query database, return data

Usage:

1. •Create Data Node
2. •Enter natural language query
3. •AI generates SQL
4. •Execute query
5. •Return data table

Example:

Input: "Query order data for the last 30 days"

Generated SQL:
SELECT * FROM orders
WHERE created_at >= DATE_SUB(NOW(), INTERVAL 30 DAY)

Output: DataFrame (order data)

Advanced Usage:

• •View generated SQL
• •Manually modify SQL
• •Set query parameters
• •Limit return rows

2. Chart Node

Function: Visualize data as charts

Supported Chart Types:

• •Line Chart
• •Bar Chart
• •Pie Chart
• •Scatter Chart
• •Area Chart
• •Table

Usage:

1. •Create Chart Node
2. •Connect to Data Node
3. •Describe the chart you want
4. •AI automatically generates chart configuration
5. •View chart

Example:

Input: "Draw sales trend chart"

AI Configuration:
- Chart type: Line chart
- X-axis: Date
- Y-axis: Sales amount
- Title: Sales Trend

Output: Line chart

Customization Options:

• •Modify chart type
• •Adjust color theme
• •Set axis labels
• •Add data labels

3. Python Node

Function: Execute Python code, process data

Usage Scenarios:

• •Complex data processing
• •Custom calculations
• •Data cleaning
• •Statistical analysis

Usage:

1. •Create Python Node
2. •Connect to Data Node
3. •Describe the operation to execute
4. •AI generates Python code
5. •Execute code
6. •Return result

Example:

Input: "Calculate month-over-month growth rate"

Generated Code:
import pandas as pd

# Input data
df = input_data

# Group by month
monthly = df.groupby(df['date'].dt.to_period('M'))['amount'].sum()

# Calculate month-over-month growth rate
growth_rate = monthly.pct_change() * 100

# Output result
output_data = pd.DataFrame({
    'month': monthly.index.astype(str),
    'amount': monthly.values,
    'growth_rate': growth_rate.values
})

Available Libraries:

• •pandas: Data processing
• •numpy: Numerical computation
• •matplotlib: Plotting
• •scikit-learn: Machine learning

4. Excel Node

Function: Import or export Excel files

Import Excel:

1. •Create Excel Node
2. •Upload Excel file
3. •Select worksheet
4. •Preview data
5. •Import as DataFrame

Export Excel:

1. •Create Excel Node
2. •Connect to Data Node
3. •Configure export options
4. •Download Excel file

5. Web Search Node

Function: Search web information

Usage Scenarios:

• •Get real-time information
• •Supplement external data
• •Market research

Example:

Input: "Search latest iPhone 15 prices"

Output: Search result summary

---

4. Practical Cases

Case 1: Sales Data Analysis

Goal: Analyze sales for the last 3 months

Steps:

1. Create Data Node - Query Order Data

Input: "Query order data for the last 3 months, including date, amount, and region"

2. Create Chart Node - Sales Trend

Connect to Data Node
Input: "Draw daily sales trend chart"

3. Create Chart Node - Regional Distribution

Connect to Data Node
Input: "Draw pie chart of sales by region"

4. Create Python Node - Calculate Growth Rate

Connect to Data Node
Input: "Calculate week-over-week growth rate"

5. Create Chart Node - Growth Rate Chart

Connect to Python Node
Input: "Draw growth rate bar chart"

Final Effect:

Data Node (Order Data)
  ├─> Chart Node (Sales Trend Chart)
  ├─> Chart Node (Regional Distribution Chart)
  └─> Python Node (Calculate Growth Rate)
        └─> Chart Node (Growth Rate Chart)

Case 2: User Behavior Analysis

Goal: Analyze user activity and retention rates

Steps:

1. Query User Login Data

Data Node: "Query user login records for the last 30 days"

2. Calculate Daily Active Users

Python Node: "Calculate daily active user count"

3. Draw DAU Trend

Chart Node: "Draw daily active user trend chart"

4. Calculate Retention Rate

Python Node: "Calculate 1-day, 7-day, and 30-day retention rates"

5. Display Retention Data

Chart Node: "Draw retention rate table"

Case 3: Financial Report Generation

Goal: Generate monthly financial report

Steps:

1. Query Revenue Data

Data Node 1: "Query this month's revenue details"

2. Query Expense Data

Data Node 2: "Query this month's expense details"

3. Merge and Calculate

Python Node: "Merge revenue and expenses, calculate net profit"

4. Generate Report

Chart Node: "Generate financial report table"

5. Export to Excel

Excel Node: "Export to Excel file"

---

5. Usage Tips

1. Node Naming

Good Naming:

• •✅ "Query Order Data"
• •✅ "Sales Trend Chart"
• •✅ "Calculate Growth Rate"

Bad Naming:

• •❌ "Data Node 1"
• •❌ "Chart"
• •❌ "Python"

2. Organizing Structure

Left-to-Right Layout:

Data Source → Data Processing → Visualization

Top-to-Bottom Grouping:

Main flow at top
Auxiliary analysis at bottom

3. Reusing Nodes

Scenario: Multiple charts use the same data

Method:

• •Create one Data Node
• •Multiple Chart Nodes connect to the same Data Node
• •Avoid duplicate queries

4. Debugging Tips

View Intermediate Results:

• •Click node to view output data
• •Check if SQL is correct
• •Verify Python code logic

Step-by-Step Execution:

• •First create Data Node, verify data
• •Then create Chart Node, verify chart
• •Finally create Python Node, verify calculation

5. Performance Optimization

Reduce Data Volume:

• •Add WHERE conditions in SQL
• •Use LIMIT to restrict rows
• •Only query needed fields

Avoid Duplicate Queries:

• •Reuse Data Node
• •Use Python Node to process data
• •Cache intermediate results

---

6. Collaboration Features

1. Real-time Collaboration

Multi-person Editing:

• •See others' cursors
• •Real-time sync of node changes
• •Avoid conflicts

2. Comments and Discussions

Add Comments:

• •Add comments on nodes
• •@mention team members
• •Discuss analysis approaches

3. Version History

View Historical Versions:

• •View modification records
• •Restore to historical versions
• •Compare version differences

4. Sharing and Export

Share Canvas:

• •Generate share link
• •Set access permissions
• •Embed in other systems

Export:

• •Export as PDF
• •Export as image
• •Export data as Excel

---

7. Common Problems

Problem 1: Node Execution Failed

Causes:

• •SQL syntax error
• •Data doesn't exist
• •Insufficient permissions
• •Timeout

Solutions:

• •View error message
• •Check SQL statement
• •Verify datasource connection
• •Increase timeout

Problem 2: Chart Display Abnormal

Causes:

• •Incorrect data format
• •Field type mismatch
• •Data is empty

Solutions:

• •Check data format
• •Convert field types
• •Add data validation

Problem 3: Python Code Error

Causes:

• •Syntax error
• •Library doesn't exist
• •Data type error

Solutions:

• •Check code syntax
• •Use supported libraries
• •Add type conversion

---

8. Best Practices

1. Canvas Design Principles

Clear Structure:

• •Left-to-right data flow
• •Put related nodes together
• •Use comments to explain

Reasonable Granularity:

• •One node does one thing
• •Avoid overly complex nodes
• •Properly split large nodes

Maintainability:

• •Use meaningful naming
• •Add necessary comments
• •Keep code concise

2. Performance Optimization

Data Query:

• •Only query needed data
• •Use indexes to speed up queries
• •Avoid full table scans

Node Execution:

• •Execute independent nodes in parallel
• •Reuse intermediate results
• •Cache commonly used data

3. Team Collaboration

Naming Convention:

• •Unified naming style
• •Clear node descriptions
• •Standard code format

Documentation:

• •Add Canvas descriptions
• •Record key logic
• •Update change logs

---

9. Summary

Canvas is AskTable's core feature, providing:

Core Capabilities: ✅ Non-linear thinking modeling ✅ Visual dependency relationships ✅ Multiple node types ✅ Real-time collaborative editing

Usage Scenarios: ✅ Complex data analysis ✅ Multi-dimensional comparison ✅ Data processing workflows ✅ Team collaborative analysis

Best Practices: ✅ Clear structural design ✅ Reasonable node granularity ✅ Performance optimization ✅ Team collaboration standards

Next Steps:

• •Canvas Advanced Tips
• •AskTable AI Agent Working Principle

---

Related Reading:

• •AskTable Canvas Real-time Collaboration Architecture
• •AskTable Canvas Streaming Execution and Breakpoint Resume

Technical Exchange:

• •Website: asktable.com