The VegaFusionWidget provides configurable debouncing to control how frequently user interaction updates are sent from the browser to the Python kernel.

The widget’s debounce_wait and debounce_max_wait properties correspond to the wait and max_wait properties of the lodash debounce function. VegaFusion uses sensible defaults (debounce_wait of 30ms and debounce_max_wait of 60ms), but it can be useful to increase the default values to handle interactions on large datasets.

Here is the specification for a chart that provides histogram data selection on a 10 million row dataset.

import altair as alt
from vega_datasets import data
import pandas as pd
import vegafusion as vf

source = pd.concat([data.flights_2k()] * 5000).reset_index()

brush = alt.selection(type='interval', encodings=['x'])

# Define the base chart, with the common parts of the
# background and highlights
base = alt.Chart().mark_bar().encode(
    x=alt.X(alt.repeat('column'), type='quantitative', bin=alt.Bin(maxbins=20)),

# gray background with selection
background = base.encode(

# blue highlights on the transformed data
highlight = base.transform_filter(brush)

# layer the two charts & repeat
chart = alt.layer(
).repeat(column=["distance", "delay"])

widget = vf.jupyter.VegaFusionWidget(chart)

Default debouncing

Here is what the default interactivity looks like using on a 2015 Macbook pro.

You can see that the server falls behind, resulting in a poor interaction experience.

Slower debouncing

Debouncing can be adjusted to improve the interaction experience. By setting debounce_wait to 50ms and debounce_max_wait to None the client will not send a request to the server until there has been a period of 50ms without any user interaction events. This results in a user interaction experience where the chart is not updated until the full interaction has completed.

vf.jupyter.VegaFusionWidget(chart, debounce_wait=50, debounce_max_wait=None)