data-visualization

You can generate publication-quality charts from cuDF outputs by applying this data-visualization Skill, which configures matplotlib for headless rendering and saves high-resolution PNG files directly to /workspace/ for inline display. It supports bar, line, scatter, and heatmap visualizations from numerical datasets.

You can build multi-panel visual summaries combining one to four charts per figure, including feature-importance plots, cluster visualizations, and confusion matrices. These publication-quality multi-panel figures are designed for clear communication of statistical findings and model results in reports or presentations.

Yes, you can visualize cuML model results such as confusion matrices without manual setup. The Skill directly processes model outputs to produce print-quality styled charts, eliminating the need for trial-and-error layout tuning or manual matplotlib configuration.

Matplotlib supports headless rendering in a GPU environment by configuring the Agg backend. This approach allows the Skill to render and save high-resolution PNG visualization files to the /workspace/ directory, requiring a read_file call to display the saved images inline.

The best way to visualize numerical analysis outputs as boxplots and histograms is using this data-visualization Skill, which applies consistent print-quality styling to statistical distribution charts. It directly accepts numerical datasets and analysis results to generate formatted visual summaries.