seaborn

Use sns.heatmap() with a 2D DataFrame or array, setting annot=True and fmt to display cell values. For correlation matrices, pass df.corr() with cmap='coolwarm' and center=0 for a diverging colormap centered on zero.

Axes-level functions like scatterplot and boxplot draw on a single matplotlib Axes and accept an ax parameter. Figure-level functions like relplot, displot, and catplot manage entire figures with built-in faceting via col and row parameters.

Seaborn is built on matplotlib and suits statistical plots with DataFrames, automatic aggregation, and attractive defaults. Use matplotlib directly for low-level control, then combine both since seaborn axes accept standard matplotlib customization.

Seaborn prefers long-form tidy data where each variable is a column and each observation is a row. Wide-form data works for heatmaps and simple plots, and df.melt() converts wide data to long format.

The KDE bandwidth controls smoothness. Adjust it with the bw_adjust parameter: values below 1 produce a less smooth curve, while values above 1 produce a smoother estimate.

Use seaborn.objects for complex layered visualizations, fine-grained control over transformations, and programmatic plot generation. The function interface remains better for quick exploratory analysis and single-purpose plots.