Scatter Plots

Complete reference for scatter plot visualizations in RING-5.

Overview

Scatter plots visualize relationships between two continuous variables. Essential for correlation analysis and outlier detection.

Configuration

Required:

• X-axis: First variable (cache_miss_rate)
• Y-axis: Second variable (ipc)

Optional:

• Color by: Category (config, benchmark)
• Size by: Third variable (execution_time)
• Shape by: Another category

Use Cases

Correlation Analysis

Identify relationships between metrics:

X: cache_miss_rate
Y: ipc
Color by: config
# Shows IPC vs miss rate correlation

Outlier Detection

Find anomalous data points:

X: memory_bandwidth
Y: execution_time
Color by: benchmark
# Identifies outliers

Multi-Dimensional Visualization

Show three dimensions:

X: cache_misses
Y: ipc
Size by: instruction_count
Color by: config
# Three metrics in one plot

Styling

Marker Properties

• Size: 6-10px (or variable)
• Opacity: 0.6-0.8 for overlapping points
• Shape: Circle (default), square, triangle

Colors

• By category: Distinct colors
• By value: Sequential or diverging scale
• Transparency for density

Data Preparation

Filtering

Remove outliers if needed:

{
    "type": "conditionSelector",
    "column": "ipc",
    "mode": "between",
    "min": 0.5,
    "max": 10.0
}

Normalization

Scale variables for comparison:

{
    "type": "normalize",
    "normalizeVars": ["x_var", "y_var"]
}

Best Practices

1. Use transparency for dense data
2. Add trend lines for correlation
3. Limit color categories to 5-7
4. Use log scale for wide ranges
5. Annotate key points

Common Patterns

Performance Correlation

X: metric_1
Y: metric_2
Color by: config
Pipeline: Filter outliers
Styling: Add trend line

Benchmark Clustering

X: ipc
Y: cache_miss_rate
Color by: benchmark_suite
Pipeline: Normalize

Next Steps

• Histograms: ../histogram-plot.md
• Line Plots: Line-Plots.md