matplotlib

Use plt.subplots(nrows, ncols) to create a figure with a grid of axes, then index into the returned axes array. For irregular layouts, use plt.subplot_mosaic or GridSpec to span rows and columns.

The object-oriented interface (fig, ax = plt.subplots()) is recommended for most code because it gives explicit control over figures and axes. Reserve the pyplot state-machine interface for quick interactive exploration.

Use matplotlib for fine-grained control over every plot element and custom plot types. Use seaborn for quick statistical plots and plotly for interactive visualizations; both integrate with or complement matplotlib.

Call plt.savefig with dpi=300 and bbox_inches='tight' for publication-quality PNG output. Use PDF or SVG formats for scalable vector graphics in print publications.

Overlapping happens when the default layout lacks spacing. Create the figure with constrained_layout=True, call tight_layout(), or save with bbox_inches='tight' to include all elements.

Use perceptually uniform sequential colormaps like viridis or cividis for ordered data, diverging maps like coolwarm for data centered on zero, and qualitative maps like tab10 for categories. Avoid jet, which is not perceptually uniform.