Zoom Logo

Workshop/Tutorial I - 14:00 - 16:00 pm UTC - Shared screen with speaker view
Sylvain Corlay
00:44
EVERYONE MUTE
Reshama S (NYC)
00:59
Hi Sylvain!
PyData Global Workshop/Tutorial I
01:31
Hello everyone! Please make sure you are muted when not speaking. Thank you!
Nicolas Kruchten
02:16
could the host please mute everyone?
Nicolas Kruchten
02:16
🙂
jackb
02:18
Can the host mute everyone?
Marc - Data Ninja
07:49
Actually I am now a Lead Trading Analysts. But same same :-)
Marc - Data Ninja
21:57
Could the moderator please remind me 1 min before my time runs out? Thanks.
Darron Fuller
23:11
Are versions of the Dash backends for the various language platforms released simultaneously? If not, what is the typical release lag between platform backends?
Erik Mason
24:10
that was great, Nicolas! Thank you! big plotly fan here :)
ARosa
24:22
I am not a front-end developer so not sure if my questions makes sense or not but I would like to know if there is any limitation in the number of users that can at the same time simultaneously interact with the dash app. Do you have any volume metric to share?
Will Patton
24:30
I cannot thank the Plotly team enough for html5 export. It's been a lifesaver for clients without infrastructure and money for BI platforms!
stanislawrobak
25:00
Can you post here link to dash presentation?
Nicolas Kruchten
25:29
Here is a link to my slides: https://bit.ly/pdg-dash
Matthew Widjaja
26:00
For those like me who was typing the GitHub link to his Panel repo: https://github.com/marcskovmadsen/awesome-panel-introduction
Erik Mason
26:16
thank you Matthew Widjaja!
Nicolas Kruchten
28:54
@ARosa There is no limit to the number of users that are using a Dash app simultaneously. Dash apps use up server resources only while executing callbacks
Nicolas Kruchten
29:34
so you can have hundreds of users looking at an app at any given time, and the server only feels load when they interact with the app. If the server is too-heavily loaded, then it's time to add servers, and that's basically trivial given the stateless architecture
Nicolas Kruchten
30:25
@Darron Fuller the Python version of the framework definitely leads but we try to push out features to the other language backends fairly promptly 🙂
James Bednar (he/him), Speaker
30:47
Each of the other frameworks is different in that respect. E.g. Panel *can* be stateless, if you deliberately program it that way, but won’t be by default. Voila and Streamlit are the opposite, in most ways, easily making all state available to the programmer but thus more costly per user.
James Bednar (he/him), Speaker
31:24
See https://panel.holoviz.org/about/comparisons.html for Panel’s viewpoint on such differences.
Radovan Kavicky
34:46
@James Bednar + all speakers in this panel... thank you for this/this is 100% awesome-ness (all the advantages of Tableau but in Python/Flask with React as frontend... and ability to develop not only Python also in Julia... absolutely perfect!/groundbreaking)
Reshama S (NYC)
35:08
+1
Nicolas Kruchten
35:37
Dash also plays nicely with Datashader and Dask! Here's an app that uses both: https://dash.gallery/dash-world-cell-towers/
Nicolas Kruchten
36:05
And we have a nice HoloViews integration as well: https://dash.plotly.com/holoviews
James Bednar (he/him), Speaker
36:53
Yep, we have lots of cool cross-pollination and integrations now! E.g. Panel apps can be deployed with Voila, as ipywidgets!
James Bednar (he/him), Speaker
37:17
And of course HoloViews supports Plotly.
Radovan Kavicky
39:26
Just wow guys... can't wait to get my hands on all of this new stuff...
Erik Mason
40:00
Thank you Marc Skov Madsen! super great presentation!
Will Patton
40:27
amazing
Nicolas Kruchten
40:35
Nice job Marc! Thanks 🙂
Reshama S (NYC)
40:55
Yes
Jorge
40:57
yes
Nicolas Kruchten
43:59
I <3 Jupyter!
Nicolas Kruchten
44:56
There is a FigureWidget yes! Any Plotly figure can be trivially upgraded to be a widget
Nicolas Kruchten
45:08
https://plotly.com/python/figurewidget/
Nicolas Kruchten
47:01
Here's the link to that great post: https://blog.jupyter.org/looking-at-notebooks-from-a-new-perspective-bfd06797f188
Erik Mason
54:13
I have to jump to a meeting, will the recordings be available to participants?
James Bednar (he/him), Speaker
54:32
Yes, I believe so!
Adam Kinsey (He)
54:37
Yes, the recordings will be made available within 1 day.
Bonora Groome, Giordano
54:51
Merci Sylvain!
Adam Kinsey (He)
54:53
I believe via YouTube.
Erik Mason
55:03
Thanks James! it's been so great! so thankful for all the presenters and the companies they represent!
Nicolas Kruchten
55:04
Such impressive work, thanks Sylvain!
Erik Mason
55:14
thanks Sylvain!
Erik Mason
55:30
and ahead of time, thanks Adrien Treuille! Love Streamlit!
Marc - Data Ninja
55:30
If you want to experience with Panel + Pyiodide checkout https://github.com/marcskovmadsen/panel-sketch. Also works with Transcrypt (python to javascript transpiling)
Marc - Data Ninja
56:56
More hair back then at PyData 2019 Adrian Treuille :-)
Fanele Chester
58:51
Looks so cool 🙌🏾
raj
59:19
cool video
James Bednar (he/him), Speaker
59:24
Look who has a marketing department! :-)
Marisa Smith
59:24
Streamlit 1.0 launch video on Youtube if you want to check it out again later: https://www.youtube.com/watch?v=c9k8K1eII4g&t=8s
Bonora Groome, Giordano
59:27
Nice!
Marisa Smith
01:00:54
Project Strike a Pose video: https://www.youtube.com/watch?v=J60AgPHpOBo
raj
01:01:08
so streamlit is similar to gradio ?
raj
01:02:58
👍wow very cool
Marc - Data Ninja
01:03:35
Panel Hello World App https://www.youtube.com/watch?v=T1B30VQdReg
Marc - Data Ninja
01:04:28
Panel Streaming Video https://www.youtube.com/watch?v=y-i23R1v9YI
Marisa Smith
01:05:12
Streamlit Main page: https://streamlit.io/
James Bednar (he/him), Speaker
01:06:04
I love watching the dueling links appear here. :-)
raj
01:06:04
so with streamlit I do not need to use matplotlib and seaborn?
James Bednar (he/him), Speaker
01:06:26
You can use lots of plotting libraries with streamline.
James Bednar (he/him), Speaker
01:06:33
streamlit !
Marisa Smith
01:06:34
You use the python graphing or visualization tool that your most comfortable with
Marc - Data Ninja
01:06:48
Links related to my Panel talk https://github.com/marcskovmadsen/awesome-panel-introductionhttps://panel.holoviz.org/index.htmlhttps://twitter.com/Panel_orghttps://twitter.com/Panel_orghttps://www.linkedin.com/company/panel-orghttps://awesome-panel.org/https://awesome-panel.org/awesome-listhttps://github.com/rapidsai/cuxfilterhttps://towardsdatascience.com/visualization-and-interactive-dashboard-in-python-c2f2a88b2ba3https://github.com/MarcSkovMadsen/panel-highcharts
Marc - Data Ninja
01:07:01
Please give Panel a ⭐ https://github.com/holoviz/panel
Marisa Smith
01:07:03
Ive always used Matplotlib but recently I have been switching to use Altair and poorly!
Marisa Smith
01:07:09
plotly*
Nicolas Kruchten
01:07:26
The classic autocorrect: plotly -> poorly 😛
Bastien
01:07:30
Someone needs to make a flowchart to decide which one of these awesome platform we should use in which case :)
Marisa Smith
01:07:42
Streamlit Cloud! 🥳: https://streamlit.io/cloud
raj
01:08:01
engineers at uber were using it internally for dashboarding?
Marc - Data Ninja
01:08:30
Jupyterhub + ContainDS for deploying any of these app frameworks with one click https://containds.com/dashboards/
Nicolas Kruchten
01:08:35
If folks are interested, we're running a private beta for Dash Cloud as well, since we're here ;)
Nicolas Kruchten
01:08:36
http://plotly.cloud/
James Bednar (he/him), Speaker
01:08:50
Finally there is an advantage to Panel not being web-searchable: at least there’s no issue with autocorrect!
Marisa Smith
01:09:21
Demo GitHub: https://github.com/streamlit/demo-self-driving
Marisa Smith
01:09:47
@raj: yes
Marc - Data Ninja
01:10:18
Panel live app gallery https://panel-gallery.pyviz.demo.anaconda.com/
raj
01:10:23
how I can listen more of Adrien and streamlit!! It looks so cool. I will be learning it today !
Marisa Smith
01:10:46
Gallery for you to get inspiration!: https://streamlit.io/gallery
Marisa Smith
01:11:18
A link to our components page as well, which are extra plug-ins for Streamlit that allow you to do even more: https://streamlit.io/components
raj
01:11:24
wow you made cheapsheet within streamlit :)
Marc - Data Ninja
01:11:38
Play around with Panel notebooks and apps live on Binder https://mybinder.org/v2/gh/holoviz/panel/v0.12.1?urlpath=lab/tree/examples
Marisa Smith
01:11:42
Community forum, come and join us!: https://discuss.streamlit.io/
Marisa Smith
01:11:53
We also have a discord: https://discord.com/invite/bTz5EDYh9Z
Clair Sullivan
01:12:25
Would have loved to have met up with y'all in CO!
Marisa Smith
01:12:52
@Clair 😻
Fanele Chester
01:13:03
Yay for discord!
raj
01:13:10
dumb question - hosting of our python app is possible on streamlit?
raj
01:13:24
like gradio
Marc - Data Ninja
01:13:40
Panel video component here https://panel.holoviz.org/reference/widgets/VideoStream.html#widgets-gallery-videostream
raj
01:14:00
YES!!!!!!!!!!!!!!
Marisa Smith
01:14:09
@ raj yes that’s the cloud platform! We have a free tier that allows you to deploy 3 apps for free with 1 GB of ram and cpu each
Marisa Smith
01:14:38
Book here: https://www.amazon.ca/Getting-Started-Streamlit-Data-Science/dp/180056550X/ref=sr_1_1?crid=3NTQYCM3LUMAR&dchild=1&keywords=Streamlit&qid=1635432577&sprefix=streamlit%2Caps%2C188&sr=8-1
Marc - Data Ninja
01:14:59
Panel Speech to Text widget https://panel.holoviz.org/reference/widgets/SpeechToText.html#widgets-gallery-speechtotext
raj
01:15:03
EVEN AI prefers streamlit
Wesley Lau
01:15:11
no freaking way... hahaha
raj
01:15:12
are you posting all the sliders of your presentation somewhere.
Nicolas Kruchten
01:15:30
Thanks Adrien, amazing work!
Marisa Smith
01:15:33
@raj: I will get you the slides after the talk + debate is over!
Bonora Groome, Giordano
01:15:42
Thank you Adrien! Was very cool
Wesley Lau
01:15:49
super cool, thanks
Sarah O'Neill (she/her)
01:16:09
Great presentations all, thanks!
raj
01:16:19
So are we voting for a winner?
Manuel Diaz
01:16:33
Amazing presentations! Thank you all for such brilliant work
Sarah O'Neill (she/her)
01:17:57
Jupyter is awesome
raj
01:18:08
another dumb question. Potentially streamlit can be used to make a webpage.
Marisa Smith
01:18:17
Yes of course!
Marisa Smith
01:18:47
Here is an app that a community member made that helps explain ML models: https://share.streamlit.io/nelsonchris1/ml-explainability-app/main/app.py
raj
01:19:15
Thanks Marisa.
Marisa Smith
01:19:22
No problem!
MatiasThayer
01:19:44
sorry if it's a silly question: how easy would be to embed any of these cool plots in a Django app?
James Bednar (he/him), Speaker
01:20:02
Panel’s website has a section about doing that.
MatiasThayer
01:20:33
really? that sounds great. I'll look for that
raj
01:21:31
anyone here has done their portfolio of projects website in streamlit?
Sylvain Corlay
01:22:00
and then "Voilà!"
Cameron Riddell
01:22:38
@Matias https://panel.holoviz.org/user_guide/Django_Apps.html?highlight=django
MatiasThayer
01:23:44
thanks!
Marisa Smith
01:24:38
@raj: checkout Pavan’s Github, he has many many repositories that he has turned into streamlit apps: https://github.com/Pavankunchala
Marisa Smith
01:24:55
He is one of our community members
Sylvain Corlay
01:25:08
Déjàvu
Marc - Data Ninja
01:25:20
Export and embedding of Panel. Check out https://panel.holoviz.org/user_guide/Deploy_and_Export.html
Marc - Data Ninja
01:27:25
Built in authentication in Panel https://panel.holoviz.org/user_guide/Authentication.html
Jan Eichhorn
01:28:42
I'd love to hear more about topic 12: Stateful apps, async processing (e.g. FastAPI backend)
Marc - Data Ninja
01:29:02
I also deploy Panel on Jupyterhub. But using ContainDS https://containds.com/dashboards/
James Bednar (he/him), Speaker
01:29:13
More votes here please!
Cameron Riddell
01:29:44
@James Bednar, where is the poll located? Can't seem to find it
James Bednar (he/him), Speaker
01:29:52
Just type a number here!
James Bednar (he/him), Speaker
01:30:07
(I’m collating responses)
Young, Adam
01:30:07
7
Jan Eichhorn
01:30:10
14
Alina (they/them)
01:30:13
4
Jorge
01:30:16
3.14
Alina (they/them)
01:30:21
42
antonella
01:30:21
20 and 14
raj
01:30:37
4
Cameron Riddell
01:31:06
11
Snyder, Jared
01:31:25
19, 20
Cameron Riddell
01:33:15
panel custom components https://panel.holoviz.org/user_guide/Custom_Components.html
Jan Eichhorn
01:33:47
Topic list numeration has changed??
James Bednar (he/him), Speaker
01:34:22
The screen share shows the numbers.
PyData Global Workshop/Tutorial I
01:34:33
Don’t think so. It has been the same what’s visible now.
V
01:34:54
can you reiterate over exporting to html (3) as way to "archive" interactive plots so could be public usable for decades to come. If else you se other solutions for 10 to 50 years lifetime?
Jan Eichhorn
01:34:54
I thought 12 was complex apps, but I might have been confused.
German Priks
01:35:42
What about accessibility support, like keyboard-only interactions?
Jorge
01:37:17
not in the list, but I propose discussing transitions, alas d3.
Marisa Smith
01:37:45
roadmap: https://share.streamlit.io/streamlit/roadmap/
Marisa Smith
01:39:14
Streamlit Session State launched in version 0.84: https://discuss.streamlit.io/t/version-0-84-0/14542/7
Marisa Smith
01:40:28
Here is our blog post on it that has some examples for you: https://blog.streamlit.io/session-state-for-streamlit/
Marisa Smith
01:44:16
Our very active community has made a few components and streamlit extras that can turn your app into a multipage app now: https://discuss.streamlit.io/search?q=multipage%20%23streamlit-examples
Nicolas Kruchten
01:45:08
How could I forget to mention Vaex! We <3 Vaex
Marisa Smith
01:46:59
Our api reference link to different progress and status updates available: https://docs.streamlit.io/library/api-reference/status
German Priks
01:47:24
Thanks, Nicolas! :)
Marisa Smith
01:47:38
st.balloons is my personal fav 🎈💕
Marc - Data Ninja
01:47:51
Panels progress indicators https://panel.holoviz.org/reference/index.html#indicators
Marisa Smith
01:51:20
Streamlit Github, ⭐us to follow along with all the new updates: https://github.com/streamlit/streamlit
Caner Bas (he/him), Istanbul
01:52:09
great one ! thanks a lot all
Jessica Walkenhorst
01:52:12
Does anybody have experience with comparing authentification workflows for the apps? I have tried 3 out of 4 and never found a good way to integrate with an auth workflow
Marc - Data Ninja
01:52:29
Panel Github ⭐ https://github.com/holoviz/panel. See you on Discourse https://discourse.holoviz.org/.