Building a dashboard in Python#

In a previous guide we discovered how we could build data pipelines in Python, here we will pick up where we left of and build an entire dashboard in Python.

To start with let us declare the Pipeline we will be working with again and initialize the Panel extension so we can render output and tables.

import lumen as lm
import panel as pn


pipeline = lm.Pipeline.from_spec({
    'source': {
        'type': 'file',
        'tables': {
            'penguins': ''
    'filters': [
        {'type': 'widget', 'field': 'island'},
        {'type': 'widget', 'field': 'sex'},
        {'type': 'widget', 'field': 'year'}
    'auto_update': False


Attaching Views#

Attaching a View to a Pipeline only requires passing the pipeline as an argument to the View constructor. The View will now be linked to the pipeline and update when we change it:

scatter = lm.views.hvPlotView(
    pipeline=pipeline, kind='scatter', x='bill_length_mm', y='bill_depth_mm', by='species',
    height=300, responsive=True


Now let us create one more view, a Table:

table = lm.views.Table(pipeline=pipeline, page_size=10, sizing_mode='stretch_width')


Laying out views#

Now we could lay these components out using Panel and publish a Panel dashboard but for now we will stick entirely with Lumen components. The Lumen Layout component will let us arrange one or more views. Here we will take our two views and put the in layout:

layout = lm.Layout(views={'scatter': scatter, 'table': table}, title='Palmer Penguins')


Building the dashboard#

Finally we can add our Layout to a Dashboard instance and give the dashboard a title via the config option.

lm.Dashboard(config={'title': 'Palmer Penguins'}, layouts=[layout])


A Dashboard (like most other components) can be previewed by displaying itself in notebook environments or by using .show() in a REPL. To serve it as a standalone application use the .servable() method and launch the notebook or script with panel serve