Plotting¶

PlotModel draws a model's network with Graphviz. It needs the viz extra and the Graphviz programs; see Inspecting a model.

PlotModel ¶

PlotModel(model: object, filename: str | Path, plot_neuron_name: bool = False, view: bool = False, title: str | None = 'Self-organizing deep learning neural network')

Plot self-organizing neural network (multilayered GMDH)

Parameters:

Name	Type	Description	Default
`model`	`SONN`	The model to draw.	required
`filename`	`str or Path`	Where `plot` writes, without the extension: the drawing goes to `<filename>.svg` and its Graphviz source to `<filename>`.	required
`plot_neuron_name`	`bool`	Write each neuron's family into its box.	`False`
`view`	`bool`	Open the drawing once it is written.	`False`
`title`	`str or None`	The line at the top of the drawing. None or an empty string leaves it out.	`"Self-organizing deep learning neural network"`

title `instance-attribute` ¶

title = title

custom palette¶

self.node_color = '#ccd1ff' self.io_node_color = '#bed4eb' self.io_pen_color = '#535ebe' self.pen_color = '#535ebe' self.io_font_color = 'black' self.connection_color = '#ccd1ff' self.connection_fill_color = '#535ebe'

add_connection ¶

add_connection(layers: Sequence[object], neuron: object, neuron_idx: int, u_index: int) -> None

Add the edge into neuron_idx of neuron from its input u_index.

The source node is an input feature or a neuron of an earlier layer, as resolved from the neuron's input layout.

add_edge ¶

add_edge(a: str, b: str) -> None

Add a directed edge from node a to node b.

plot ¶

plot() -> None

Build the graph and render it to <filename>.svg.

Draws the input features, every neuron of every layer, their input edges and the readout (best neuron or out_proj head). Does nothing for a model with no layers.

Raises:

Type	Description
`RuntimeError`	If the Graphviz `dot` executable is not on PATH.