A GIF demoing interaction with an ipympl Matplotlib figure. In the demo,
Clicking a cluster opens a pop-up window with a new figure.
The background Matplotlib figure shown is rendered as an interactive ipympl
widget and wrapped in IPyOverlay’s
OverlayContainer
, and shows a UMAP of
clustered embeddings. Clicking a cluster or right-clicking and selecting an
exhibit creates a click-and-draggable pop-up window containing a new
visualization that plots more details for the selected cluster. Optionally, each
pop-up window can show a line connecting the window to the center of the cluster
it is associated with in the underlying figure.
Custom actions can be added as A right-click menu to any widget. This menu is
specific to this Matplotlib UMAP and allows users to choose whether to open a
violin or distribution plot for the hovered cluster.
If multiple separate “background” widgets (Matplotlib UMAP on the left and
Plotly map on the right) are wrapped in an
OverlayContainer
, overlay widgets
can be displayed and dragged over all of them.