Bokeh / DASK / Holoview


This is user-contributed content. Credits go to Dr. A. Khalatyan (researcher at IT-eScience, AIP, Potsdam)!

Bokeh / DASK server

The combination of the Bokeh, DASK and Holoview is allowing us to interactively render large datasets. This example is deploying an interactive visualization of a DASK dataframe via a Holoview application:

  1. create a file named

    import os
    import dask.dataframe as dd
    import holoviews as hv
    import pandas as pd
    import numpy as np
    from holoviews.operation.datashader import datashade
    # 1. Load data
    ## load from a parquet file:
    #ddf = dd.read_parquet("data.parq")[['xgal', 'ygal']].persist()
    # or let's just generate some random data for demo
    df = pd.Dataframe({
        'xgal': np.random.randn(1000),
        'ygal': np.random.randn(1000)
    ddf = dd.from_pandas(df)
    # 2. Datashade it
    points = hv.Points(ddf, kdims=['xgal', 'ygal'])
    shaded = datashade(points).opts(plot=dict(width=800, height=600))
    # 3. Instead of Jupyter's automatic rich display, render the object as a bokeh document
    doc = hv.renderer('bokeh').server_doc(shaded)
    doc.title = 'HoloViews Bokeh App'
  2. in terminal start the Bokeh server:

    bokeh serve --disable-index-redirect --allow-websocket-origin='*' --port 5006
  3. now your interactive plot is accessible on following URL:[PROJECT_ID]/server/5006/server_app

    or run this line in a terminal to get the URL and click on it:

    echo "$COCALC_PROJECT_ID/server/5006/server_app"

Holoview App

For deploying an interactive plots based on Bokeh, Panel, Holoviz or similar libraries we can use the server feature of CoCalc, like in the examples for the Tensorboard:

  1. Create a python file (If you are using a self hosted instance, replace with your domain):

    import holoviews as hv
    import panel as pn
    import numpy as np
    def sine(frequency, phase, amplitude):
        xs = np.linspace(0, np.pi*4)
        return hv.Curve((xs, np.sin(frequency*xs+phase)*amplitude)).options(width=800)
    if __name__ == '__main__':
        ranges = dict(frequency=(1, 5), phase=(-np.pi, np.pi), amplitude=(-2, 2), y=(-2, 2))
        dmap = hv.DynamicMap(sine, kdims=['frequency', 'phase', 'amplitude']).redim.range(**ranges)
        pn.serve(dmap, port=8006, allow_websocket_origin=[""], show=False)
  2. Start it in the terminal:

  3. now your interactive Holoview application is available in the following URL:[PROJECT_ID]/server/8006/

    or run this line in a terminal to get the URL and click on it:

    echo "$COCALC_PROJECT_ID/server/8006/"