Plotly's built-in lasso selection is non-trivial to achieve in Vega-Altair.

Plotly offers more built-in chart types and supports some more exotic coordinate systems than Vega-Altair out of the box.

Since both libraries use D3 and SVG under the hood, charts can easily be converted from one library to the other.

Linking charts that use a common data attribute is more straightforward in Vega-Altair.

Vega-Altair charts are fully interactive by default.

Vega-Altair is built on top of the Vega-Lite grammar.

Vega-Altair exposes low-level graphical primitives such as patches and artists.

Vega-Altair is a declarative visualization library for Python.

.on_selection(...) allows us to handle selections on the scatter trace.

Our selection_function updates the text in an HTML widget.

FigureWidget is a built-in type of Jupyter widget.

fig.data[-1] is the last item in the plotted iris dataset.

Batching updates can prevent flickering when a Plotly widget depends on another widget.

Redrawing issues can be fixed by creating a dedicated figure widget rather than a standard Plotly figure.

Competing updates triggered by multiple widgets at similar times can cause flickering.

Creating a standard Plotly figure and placing it inside an @interactive function can cause the entire figure to be redrawn or rerendered.

Brushing in a scatterplot can be easily achieved using Jupyter widgets and seaborn.

Jupyter widgets provide interactive browser controls for Jupyter notebooks.

Jupyter widgets are incompatible with static plotting libraries like Matplotlib.

The interact function can produce different control elements for different types of arguments, even when the exact widget is not specified.

pointplot

scatterplot

swarmplot

lineplot

boxplot

kdeplot

stripplot

rugplot

Figures can be exported as raster graphics, but only in the lossy JPEG format.

One limitation of matplotlib is that figures can only be exported as raster graphics.

Figures can be readily exported as vector graphics and opened later with SVG editors.

Exporting figures as vector graphics is possible, but requires an additional library.

Seaborn is incomatible with Matplotlib.

Matplotlib provides a rich set of interactions like brushing and navigation.

Seaborn provides high-level methods for creating statistical charts.

Seaborn is built on top of Matplotlib.