Parallel Coordinates¶
The Parallel Coordinates page provides a powerful visualization for exploring high-dimensional optimization data. Each solution is represented as a line connecting values across multiple parallel axes.
Overview¶
Parallel coordinates plots are ideal for: - Visualizing many variables simultaneously - Identifying patterns and correlations - Interactive filtering through brushing - Comparing multiple solutions - Exploring trade-offs between objectives
Features¶
Result Selection¶
Select Result: Choose which optimization result to visualize - Currently supports single result selection - The plot updates automatically when you change the selection
Column Selection¶
Select Columns to Display: Choose which variables to show as axes
Available Column Types: - Design Variables: Input parameters - Objectives: Optimization goals - Inequality Constraints: Constraint values - Equality Constraints: Constraint values - Observables: Additional computed values
Default: The first 5 columns are selected automatically
Tips: - Start with objectives to see solution quality - Add key design variables to understand trade-offs - Include constraints to identify feasibility boundaries - Limit to 5-10 axes for best readability
Color Coding¶
Color By: Choose how to color the lines
Options: - Result: Color by result name (useful when comparing multiple results) - Any Column: Color by the values of a specific variable/objective - Set Membership: Color by which set solutions belong to
Benefits: - Quickly identify patterns - Highlight specific solution characteristics - Distinguish between different groups
Interactive Brushing¶
The most powerful feature of parallel coordinates is brushing:
- Click and drag on any axis to select a range
- Multiple brushes: Create ranges on multiple axes simultaneously
- Filtering: Only lines passing through all brushes are highlighted
- Reset: Click outside the brush to remove it
Use Cases: - Find solutions within specific variable ranges - Identify feasible solutions (constraint values ≤ 0) - Explore trade-off regions - Isolate interesting solution clusters
Interpretation¶
Reading the Plot¶
- Each Line: Represents one solution
- Each Axis: Represents one variable/objective
- Line Position: Shows the value on each axis
- Line Color: Indicates grouping or value (based on color selection)
Patterns to Look For¶
Parallel Lines: - Strong correlation between adjacent axes - Solutions with similar characteristics
Crossing Lines: - Negative correlation - Trade-offs between objectives
Clusters: - Groups of similar solutions - Distinct solution families
Outliers: - Unusual or extreme solutions - Potential errors or interesting edge cases
Usage Workflows¶
Workflow 1: Explore Pareto Front¶
- Select your result
- Choose all objectives as columns
- Color by "pareto" set (if created)
- Brush to explore different regions of the front
- Identify trade-off patterns
Workflow 2: Feasibility Analysis¶
- Select objectives and constraints
- Color by "feasible" set
- Brush constraint axes at ≤ 0
- See which objectives are achievable within constraints
Workflow 3: Design Space Exploration¶
- Select design variables and key objectives
- Brush objectives to select good solutions
- Observe which variable ranges produce good results
- Identify promising design regions
Workflow 4: Multi-Criteria Filtering¶
- Select relevant columns
- Brush multiple axes to define criteria
- Narrow down to solutions meeting all criteria
- Export or create a set from filtered solutions
Best Practices¶
Column Selection¶
- Start Simple: Begin with 3-5 most important columns
- Add Gradually: Add more columns as needed
- Group Related: Place related variables adjacent
- Objectives First: Start with objectives to see solution quality
Axis Ordering¶
- Place correlated variables next to each other
- Put objectives at one end for easy comparison
- Group constraints together
- Experiment with different orderings
Color Strategy¶
- Use color to highlight what matters most
- Color by objectives to see solution quality distribution
- Color by sets to compare different groups
- Color by constraints to identify feasibility
Brushing Techniques¶
- Broad to Narrow: Start with wide brushes, then refine
- One at a Time: Add brushes incrementally
- Reset Often: Clear brushes to see the full picture
- Combine: Use multiple brushes for complex filtering
Tips and Tricks¶
Performance¶
- Limit to \<5000 solutions for smooth interaction
- Use sets to pre-filter large datasets
- Reduce number of axes if plot is slow
Visualization¶
- Adjust browser zoom for better visibility
- Use full-screen mode for complex plots
- Export plots for presentations
Analysis¶
- Look for "bundles" of lines indicating solution clusters
- Identify axes where lines spread out (high variance)
- Find axes where lines converge (low variance)
- Use brushing to validate hypotheses
Common Use Cases¶
Multi-Objective Optimization¶
- Visualize Pareto front trade-offs
- Identify knee points (best compromises)
- Explore objective space structure
Constraint Handling¶
- Identify active constraints
- Find constraint boundaries
- Understand feasible region shape
Sensitivity Analysis¶
- See which variables affect objectives most
- Identify robust solution regions
- Understand parameter interactions
Solution Selection¶
- Filter to solutions meeting specific criteria
- Compare different solution families
- Narrow down to final candidates
Troubleshooting¶
Plot is Cluttered¶
- Reduce number of solutions (use sets)
- Decrease number of axes
- Use brushing to focus on specific regions
Can't See Patterns¶
- Try different color schemes
- Reorder axes
- Adjust axis scales (if available)
Brushing Not Working¶
- Ensure you're clicking and dragging on an axis
- Check that lines are visible
- Try refreshing the page
Navigation¶
- Path:
/parallel-coordinates - Category: Visualization
- Icon: Math norm icon
- Requires Data: Yes
Related Pages¶
- Interactive Scatter: 2D visualization alternative
- Scatterplot Matrix: Pairwise relationships
- Sets Manager: Create sets from filtered solutions