Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 22:12
    DSDEV27001 commented #2288
  • 20:36
    jbednar commented #2295
  • 19:27
    Hoxbro labeled #2295
  • 19:27
    Hoxbro opened #2295
  • 17:47
    nghenzi commented #2293
  • 17:44
    syamajala commented #2293
  • 17:44
    codecov[bot] commented #2198
  • 17:42
    codecov[bot] commented #2198
  • 17:39
    codecov[bot] commented #2198
  • 17:38
    codecov[bot] commented #2198
  • 17:33
    codecov[bot] commented #2198
  • 17:33
    codecov[bot] commented #2198
  • 17:32
    nghenzi commented #2293
  • 17:30
    codecov[bot] commented #2198
  • 17:30
    nghenzi commented #2293
  • 17:14
    MarcSkovMadsen closed #2294
  • 17:14
    MarcSkovMadsen commented #2294
  • 17:13
    codecov[bot] commented #2198
  • 17:13
    MarcSkovMadsen synchronize #2198
  • 17:13

    MarcSkovMadsen on binder

    Fix #2294: add requirements bac… (compare)

epifanio
@epifanio
they are CTD observation, weather station, soon permafrost - but we also have to integrate gridded data (sentinel satellite)
Philipp Rudiger
@philippjfr
Okay for gridded data you could also convert to cupy (I think xarray supports those now) and then use a HoloViews QuadMesh + datashading to render it.
Would recommend https://hvplot.holoviz.org/ for plotting, at least in the exploratory phase.
I don't know how large any of your data is so not sure how much benefit you are going to see from datashading or if you even need GPU datashading.
But hvPlot/HoloViews will transparently work with either cuDF or regular dataframes and xarrays backed by NumPy, Dask or Cupy.
epifanio
@epifanio
For gridded data I started with the landsat example which is using the xarray_raster capabilities, so I guess I will first use xarray_raster to read the data in and then convert it to a cupy array which should be understood by the quadmesh api
each nectdf data (for the gridded data) is ~2GB
James A. Bednar
@jbednar
Sounds good.
Thomas Diederen
@Patrickens_gitlab
is it possible to label datapoints in hv.Scatter? In the following example I would like a legend where the datapoints are labelled 'a' and 'b' (corresponding to color):
import holoviews as hv
import pandas as pd
import numpy as np
import random

randhex = lambda : format(random.randint(0, 255), '02x')
randcol = lambda : f'#{randhex()}{randhex()}{randhex()}'
data = pd.DataFrame({
    'x': np.arange(0,1,0.1),
    'y': np.arange(0,1,0.1),
    'color': [randcol()]*5 + [randcol()]*5,
    'label': ['a']*5 + ['b']*5,
})
hv.Scatter(data, kdims=['x', 'y'], vdims=['color', 'label'], label='label').opts(color='color', show_legend=True)
Thomas Diederen
@Patrickens_gitlab
but thats for labels in the plot, I am looking for 2 legend entries: 'a' and 'b' corresponding to the colors
and thank you for the answer!
James A. Bednar
@jbednar
You just want to turn on a legend? I forget the name, but there's an option for that!
Thomas Diederen
@Patrickens_gitlab
I looked through help(opts.Scatter) but could not find an option that I would think could do that
Thomas Diederen
@Patrickens_gitlab
should I think about something like: categorical_points.sort('Category').options(color='Category', cmap='Category20', size=5)? This is straight from the documentation, but I cannot get it to play nicely with my dataframe
James A. Bednar
@jbednar
Ah, I wasn't paying close attention to what you are doing. A legend will be generated for an overlay of points on points, to distinguish between two different sets of datapoints. So if you separate your points into two different Scatter elements then overlay them, you'll get the legend you want. You can use a groupby to set up such categories. But if you have a column with your color, you could supply an infinite number of different 'label' arguments, which isn't something legends are set up to handle, so there's no direct way to set up a legend for that case that I know of. So, first get a finite set of different groups, then plot those as an overlay.
Thomas Diederen
@Patrickens_gitlab
ok, ill post a solution in a bit to see whether Ive understood what you mean. Thank you!

as such:

randcol = lambda : f'#{randhex()}{randhex()}{randhex()}'

data = pd.DataFrame({
    'x': np.arange(0,1,0.1),
    'y': np.arange(0,1,0.1),
    'color': [randcol()]*5 + [randcol()]*5,
    'label': ['a']*5 + ['b']*5,
})
overlay = []
for label, df in data.groupby('label'):
    scatter = hv.Scatter(df, kdims=['x', 'y'], vdims=['color', 'label'], label=label).opts(color='color', show_legend=True)
    overlay.append(scatter)
hv.Overlay(overlay)

???

James A. Bednar
@jbednar
hv.Scatter(data, kdims=['x', 'y'], vdims=['color', 'label'], label='label').groupby('color').overlay()
Or, to get better labels, hv.Scatter(data, kdims=['x', 'y'], vdims=['color', 'label'], label='label').groupby('label').overlay()
Thomas Diederen
@Patrickens_gitlab
Thank you! Thats the brevity I was looking for
James A. Bednar
@jbednar
A PR to add that sort of example to the Scatter reference gallery notebook would be appreciated. :-)
You surely aren't the only one to wonder!
Thomas Diederen
@Patrickens_gitlab
@jbednar Ive never done such a thing, but this might be a good place to start
James A. Bednar
@jbednar
It's always a good time to start! And this would be useful; otherwise it will definitely fall off our radar.
David Saroff
@davidsaroff_gitlab
I'm trying to install holoviz in anaconda on a windows 10. When I execute
conda install -c pyviz holoviz
I get the error
ERROR conda.core.link:_execute(700): An error occurred while installing package 'defaults::rise-5.6.1-py37_1'.
David Saroff
@davidsaroff_gitlab
James A. Bednar
@jbednar
That's annoying. Seems like some issue with the Windows conda package for rise? Users don't actually need rise unless you want to do a live presentation (Jupyter used like PowerPoint), so hopefully there's some configuration you'll be able to install in any case. What happens if you specify a different RISE version, e.g. conda install -c pyviz holoviz rise=5.5?
geoHeil
@geoHeil
May I ask for help getting started with hvplot https://stackoverflow.com/questions/64374829/convert-matplotlib-to-interactive-holoviews-datashader-visualization-ideally when converting an existing matplotlib visualization of timeseries to an interactive data-shadered hvplot. I am stuck assigning the right colors for the individual categories and creating a dynamic filter (preferably with a brush)
Marc Skov Madsen
@MarcSkovMadsen
Hi @geoHeil . Tru
Hi @geoHeil . Try asking for help on Discourse. Please include a minimum, reproducible code example that the community can help you improve. Thanks.
Marc Skov Madsen
@MarcSkovMadsen
If you missed it Streamlit Sharing has been released. One click deployment and free hosting of open source, community apps. This is such a competitive advantage for Streamlit. I've seen Plotly tweet that they have had a description for free hosting on Heroku. But this is still too advanced for the many users of analytics apps frameworks. It can be costly for Streamlit to host. But they get data on all the users and use cases of Streamlit. It is such a brilliant move. Great for Streamlit. Great for the Streamlit community. Great for Python. https://www.streamlit.io/sharing.
Does Anaconda or someone else close to the HoloViz ecosystem see something similar as a huge opportunity to provide?
Also try clicking some of the apps to get a feel for the look and feel that Streamlit provides.
Marc Skov Madsen
@MarcSkovMadsen
streamlit_app.png
Marc Skov Madsen
@MarcSkovMadsen
At least providing something even simpler based on the Heroku description here https://dash.plotly.com/deployment could be an idea and a huge win I believe.
Andrew
@IAteAnDrew1_twitter
that's really cool with streamlit
https://panel.holoviz.org/user_guide/Server_Deployment.html I think panel already does show how to deploy on Heroku here
Kim Pevey
@kcpevey

Anyone else have issues with bokeh after creating a fresh env (as of yesterday)? I'm getting

(unknown) [bokeh] JS/Python version mismatch
(unknown) [bokeh] Library versions: JS (2.2.1) / Python (2.2.3)

Apologies if this is the wrong place, but the Bokeh gitter doesn't seem to be a thing anymore.

I don't know where/what needs to be updated to update the JS library
James A. Bednar
@jbednar
Maybe quit your browser and reload? Not sure.
@MarcSkovMadsen , hosting public apps in that way has been discussed as a possibility by Anaconda for the entire time I've been working there, but nothing concrete has happened on that front.
I agree (and have argued internally) that it would be a great idea.
Marc Skov Madsen
@MarcSkovMadsen
Sounds great @jbednar . I think now you have more arguments: 1) Streamlit does it with huge success 2) HoloViz including Panel is NOW so mature that there are lots of business opportunities for Anaconda in promoting it.
James A. Bednar
@jbednar
How do you know it's a success (financially) for streamlit?
Marc Skov Madsen
@MarcSkovMadsen
I don't :-) But I can see community wise and communication wise that it is.
James A. Bednar
@jbednar
That part has never been in doubt. :-)
Marc Skov Madsen
@MarcSkovMadsen
My understanding is that solving the deployment problem is the hard problem. And Streamlit seems to have solved it. I claim people will use their technology and stay with it just for that reason alone.
The thing I don't understand is why Heroku, AWS, Azure, Google Cloud and other cloud providers don't provide the same simple 3 step technical solution.