FigureWidget() update

[zxweed]

Hi, very useful project, all my career I dream about such thing. It seems that it can make plotly usable in real life, not only in the iris dataset.

Is there a way to dynamic update the resampled FigureWidget instance? For example, in the Jupyter lab:

The last cell causes an update of the data in the chart if fig is an FigureWidget instance, but does not update if the instance is a FigureResampler(go.FigureWidget())

Test case:

import numpy as np
from plotly_resampler import FigureResampler

x = np.arange(1_000_000)
noisy_sin = (3 + np.sin(x / 15000) + np.random.randn(len(x)) / 10) * x / 1_000

fig = FigureResampler(go.FigureWidget())
fig.add_scattergl(name='noisy sine', showlegend=True, x=x, y=noisy_sin)

fig.update_layout(autosize=True, height=300, template=None, legend=dict(x=0.1, y=1, orientation="h"),
                  margin=dict(l=45, r=15, b=20, t=30, pad=3))
fig.show()

# does not update chart if fig is FigureResampler instance
with fig.batch_update():
    fig.data[0].y = -fig.data[0].y

PS: It seems that resampling only works in dash, but not in jupyterlab?

[jvdd]

Hi @zxweed

Thank your for the nice words!

I see you are asking two things here; a) How to have dynamic resampling with plotly-resampler b) How to update the graph data

a) How to have dynamic resampling with plotly-resampler

To use plotly-resampler, one should perform 2 steps; https://predict-idlab.github.io/plotly-resampler/getting_started.html#how-to-use

1. wrap the plotly Figure with FigureResampler
2. call .show_dash() on the Figure

You did step 1 correctly, but called .show instead of .show_dash. If you do not perform step 2, no dynamic resampling will happen (as no underlying dahs app will get started)

PS: plotly-resampler works in jupyterlab as well. You should then call .show_dash(mode="inline") :blush:

import numpy as np; import plotly.graph_objects as go
from plotly_resampler import FigureResampler

x = np.arange(1_000_000)
noisy_sin = (3 + np.sin(x / 15000) + np.random.randn(len(x)) / 10) * x / 1_000

# Step 1
fig = FigureResampler(go.FigureWidget())
fig.add_scattergl(name='noisy sine', showlegend=True, x=x, y=noisy_sin)

fig.update_layout(autosize=True, height=300, template=None, legend=dict(x=0.1, y=1, orientation="h"),
                  margin=dict(l=45, r=15, b=20, t=30, pad=3))
# Step 2
fig.show_dash(mode='inline')  # visualizes the chart inline in your notebook

b) How to update the graph data

As for now, the large high-frequency sequence data is stored in a private variable hf_data. Just had a discussion with @jonasvdd about this, and we could support such an update by exposing the hf_data via a property. A future PR will tackle this issue.

Hope this answer already helps you!

Cheers, Jeroen

[jvdd]

Question b) is tackled in PR #46

[zxweed]

Is it really necessary to use dash server in Jupyter-only environment? For example, I can't open additional port 8050 on the server, only 8888 is available. After all, the FigureWidget has its own event handlers, on which you can hook your fast resamplers:

internal.resample.ipynb.zip

How hard is it to implement?

[jvdd]

Oh, I think I get what you mean! That is a very interesting suggestion :smiley:

I'll look into it and will keep you posted in this thread, @jonasvdd also sees the potential of supporting FigureWidget as such no port forwarding is required for the end-users (in Jupyter)

Thanks for your input!

[jonasvdd]

@zxweed, very good issue!!

I think it is feasible to implement and has a high-relevance!

However, I will need to break my head about how to implement this without losing backward compatibility & still having clean code. Will further look into it next week!

(fyi: at the end of this week, we will release a new version #46 which makes our aggregators faster and memory efficient 2x speedup) 🔥

[jonasvdd]

see #47 🙏🏼

[jonasvdd]

@zxweed PR is merged, you can try it out with v=0.6.0! Always glad to hear feedback :)

[zxweed]

Thanks, I checked in jupyter lab and notebook - works as it should, you make plotly great again :)

The parameters hf_x and hf_y are not yet available through fig.data? They cannot be changed dynamically?

[jonasvdd]

Hi @zxweed, to adjust these data properties use the hf_data property! see: https://predict-idlab.github.io/plotly-resampler/getting_started.html#dynamically-adjusting-the-scatter-data

[zxweed]

Yes, thank you, but my chart doesn't update after I change the data - I need to explicitly press some button to do that:

plotly.resampler.test.zip

[zxweed]

@jonasvdd, should I call some method to update the graph explicitly after changing the hf_data property?

[jonasvdd]

Hi @zxweed, we have a solution for your issue! will create a new release for this!

[jonasvdd]

You will indeed have to call an additional method, i.e. reset_axes on your FigureWidgetResampler

[jonasvdd]

@zxweed version 0.6.1 has the desired functionality i.e., FigureWidgetResampler.reset_axes and/or FigureWidgetResampler.reload_data

[zxweed]

Thank you, it works as it should in 0.6.1 - is this a special build for me? In the current version 0.6.2 these methods are removed...

[thanosam]

Hello,

Congratulations for such a useful library. Makes plotly so powerful. I have one question. I have an implementation in Django and using plotly (in my views), I send the plot to the frond-end(templates). I am currently using your library and I manage to plot more than 1mil datapoints in less than 1sec. (more than 15sec before !). But the dynamic aggregation doesn't work. Any ideas of how can I implement this feature in Django?

Thank you so much in advance.

[jvdd]

Hi @thanosam

To help you with your issue, I will need some more details;

• Are you working remote (i.e., creating your visualization on a server / another machine)?
• In what environment are you creating your visualization (e.g., vscode, pycharm, jupyter, or just a python script)?
• Are you using FigureResampler or FigureWidgetResampler?

Perhaps you could send some minimal code example so I can reproduce your issue on my machine?

What I can think of right now;

• You called .show() on the wrapped plotly figure
  • For FigureResampler you should call .show_dash
  • For FigureWidgetResampler you should just output the your instance in a cell
• You are working remotely and did not forward the port when using the FigureResampler

Let me know if one of the above possible issues already helped you! :)

Cheers, Jeroen

[jvdd]

Hi @zxweed

@jonasvdd created the v0.6.2 release, I suspect that he created this release from the #52 branch (which is most likely not branched from the latest version of the main branch).

We will look into it, to avoid such behavior in the next release!

Jeroen

[thanosam]

Hi @jvdd,

Thank you for your fast answer. I am working in pycharm and I use FigureWidgetResampler since I wanna avoid dash involvement.

In my current implementation I produce plots dynamically from a list. Everytime I click on an option in the list, the frond-end makes an ajax call to the back-end. Then, until today with plotly I was giving the plot back as an ajax response, adding it in a div. Now, I use your method that way. I receive the resampled plot. But when I zoom it doesn't resamples.

Example:

Frond-end -> ajax call to the back-end. In a view:

... fig = FigureWidgetResampler(go.Figure()) fig.add_trace(go.Scattergl(name='noisy sine', showlegend=True), hf_x=x, hf_y=noisy_sin) plot_div = plot(fig, output_type='div') ...

Backe-end -> sends the plot_div as a response to the ajax call and I add it in a div.

[jonasvdd]

@zxweed, again, good catch, thanks for telling us! Just deployed 0.6.3 should be fixed now! 😃

[zxweed]

Yes, it works now in the 0.6.3 version, thank you for fast fix.

[jvdd]

Hi @thanosam

I believe I discovered why no resampling occurs.

I think you use the plotly.offline.plot() function, which creates a standalone HTML. Therefore, (I think) there is no connection with the IPython environment that is required for your FigureWidgetResampler or your back-end dash application when using FigureResampler.

If possible, I would advise to check whether you can ingest the figure without using plotly.offline.plot so that the figure is not a standalone html but still connected to either your IPython environment (when using FigureWidgetResampler) or a dash backend (when using FigureResampler).

Hope this helps!

[thanosam]

Hi jvdd,

Thank you for your response. Indeed I need to use the offline version. I need to have it as an html in my response in order to change it dynamically. With this implementation there isn't something I could do?

Thanks

[jvdd]

Hi @thanosam,

I do not think there is a way to add callbacks to plotly.offline plots.

If you want adaptive resampling in your application I would suggest to use dash (as such you can have callbacks, and define your figure using plotly dash-html-componets).

More info on how you can integrate plotly-resampler in a dash application;

• our dash-app examples
• the docs about dash-app integration

Cheers

[thanosam]

Hi @jvdd ,

I am developing my app in Django and I use django-plotly-dash library for the plots. I was wondering is the plotly-resampler compatible with the django-plotly-dash library? If yes, how?

[jvdd]

Hi @thanosam,

I've created a separate issue for this (see #82) - we can further discuss it over there