Streaming Options#
Note
All the examples below require a live python process to experience the full interactivity of the plots.
Options for handling live data streams:
Parameters |
Description |
|---|---|
backlog (int, default=1000) |
Maximum number of rows to keep in the stream buffer when using a streaming data source. |
stream (holoviews.streams.Stream or None, default=None) |
A stream object for streaming plots, allowing data updates without re-rendering the entire plot. |
backlog#
The backlog option allows to define the maximum number of rows to keep in the stream buffer when using a streaming data source. Default is 1000. The object returned when stream is set is a HoloViews DynamicMap.
import hvplot.streamz # noqa
from streamz.dataframe import Random
df = Random(interval='200ms', freq='50ms')
df.hvplot.table(width=400, backlog=5)
/Users/runner/work/hvplot/hvplot/.pixi/envs/docs/lib/python3.11/site-packages/streamz/plugins.py:3: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
import pkg_resources
When running these two cells in a notebook one after the other, the data in the table above will be updated every 200ms for 2 seconds.
import asyncio
await asyncio.sleep(2)
df.stop()
stream#
The stream option accepts a stream object for streaming plots, allowing data updates without re-rendering the entire plot. The object returned when stream is set is a HoloViews DynamicMap.
In this first example, we set up and attach a HoloViews Pipe stream, we then call its send method in the next cell with a new DataFrame, which will entirely replace the plotted data, plotting a year after the other with an interval of 1 second.
import hvplot.pandas # noqa
from holoviews.streams import Pipe
df = hvplot.sampledata.stocks("pandas").set_index("date")
years = df.index.year.drop_duplicates()
sdf = df.loc[str(years[0]), :]
stream = Pipe(data=sdf)
plot_opts = dict(height=250, width=500, xlim=(df.index.min(), df.index.max()), ylim=(0, 6))
plot = sdf.hvplot.line(stream=stream, **plot_opts)
plot
import time
for year in years[1:]:
stream.send(df.loc[str(year), :])
time.sleep(1)
In this second example we set up and attach a HoloViews Buffer stream, we then call its send method in the next cell with a new DataFrame, which will accumulate the data, gradually plotting the whole dataset with an interval of 1 second.
import hvplot.pandas # noqa
from holoviews.streams import Buffer
df = hvplot.sampledata.stocks("pandas").set_index("date")
years = df.index.year.drop_duplicates()
sdf = df.loc[str(years[0]), :]
stream = Buffer(data=sdf, index=False)
plot_opts = dict(height=250, width=500, xlim=(df.index.min(), df.index.max()), ylim=(0, 6))
plot = sdf.hvplot.line(stream=stream, **plot_opts)
plot
import time
for year in years[1:]:
stream.send(df.loc[str(year), :])
time.sleep(1)