Visualization Building Blocks

Establish terminology for building figures with comparisons of matplotlib, seaborn, and plotly.

Bonus: Aviod common pitfalls

Figure

Figure: The overall container that holds everything in a visualization, including one or more plots/axes.

matplotlib

Created with plt.figure() or implicitly with plt.subplots(). Acts as the top-level container for all axes and elements.

seaborn

Wraps matplotlib; figure objects are matplotlib figures. Many seaborn functions automatically create a figure.

plotly

Created with go.Figure() or make_subplots(). Serves as the top-level object to which traces and layout are added.

Pitfalls

In matplotlib/seaborn, you usually need plt.show() in scripts to render; in notebooks it auto-displays. In plotly, you must call fig.show(). Seaborn may create its own figure silently if you don’t pass ax=.... Plotly figures are interactive by default, while matplotlib’s are static unless combined with widgets.

Axes

Axes: A single plotting area within a figure (what most people think of as ‘the plot’).

matplotlib

Created with fig.add_subplot() or plt.subplots(). Each Axes has its own coordinate system, ticks, and labels.

seaborn

Always returns a matplotlib Axes (or FacetGrid/PairGrid, which manages multiple Axes). Seaborn’s high-level API plots directly onto Axes objects.

plotly

Doesn’t use an explicit ‘Axes’ object. Instead, axes are defined by layout (fig.update_xaxes, fig.update_yaxes). Multiple subplots are managed through layout specs.

Pitfalls

Terminology confusion: ‘Axes’ (plural of ‘Axis’) vs ‘Axis’ (x/y scale). In seaborn, calling multiple functions on the same Axes requires passing ax=...; otherwise each call may create a new figure. In plotly, axis limits and labels are adjusted via layout updates, not per-object methods.

Color Palettes

Color Palette: Collection of colors used to distinguish categories or continuous values.

matplotlib

Uses ‘colormaps’ (cmap) for continuous data (e.g., viridis, plasma) and ‘color cycles’ for categorical data.

seaborn

Built-in support for categorical palettes (color_palette('deep')), sequential/diverging palettes (cubehelix_palette, light_palette), and wrappers for matplotlib colormaps.

plotly

Supports color_discrete_sequence for categorical colors and colorscale for continuous values. Many named scales available (Viridis, Cividis, Plasma).

Pitfalls

In matplotlib, cmap is only for continuous values; using it for categories requires manual mapping. Seaborn will override matplotlib’s default color cycle when you call sns.set_theme(). Plotly distinguishes between discrete (color_discrete_sequence) and continuous (colorscale); passing the wrong one has no effect.

Templates / Styles

Templates / Styles: Preset styling that controls background, grids, font, and overall look of plots.

matplotlib

Uses ‘colormaps’ (cmap) for continuous data (e.g., viridis, plasma) and ‘color cycles’ for categorical data.

seaborn

Themes via sns.set_theme(style='whitegrid'). Controls background, ticks, grid, font scale, etc.

plotly

Templates applied with fig.update_layout(template='plotly_dark'). Includes built-in templates (plotly, ggplot2, seaborn) or user-defined ones.

Pitfalls

Matplotlib stylesheets affect all following plots unless reset — beginners often forget to reset. Seaborn’s set_theme changes matplotlib defaults globally (grids, ticks, etc.), which may surprise users mixing raw matplotlib code. Plotly templates override both colors and layout; you may need to re-apply custom settings after applying a template.