facebookresearch / hiplot

HiPlot makes understanding high dimensional data easy

HiPlot - High dimensional Interactive Plotting

HiPlot is a lightweight interactive visualization tool to help AI researchers discover correlations and patterns in high-dimensional data using parallel plots and other graphical ways to represent information.

Try a demo now with sweep data or upload your CSV or

There are several modes to HiPlot:

• As a web-server (if your data is a CSV for instance)
• In a jupyter notebook (to visualize python data), or in Streamlit apps
• In CLI to render standalone HTML

pip install -U hiplot  # Or for conda users: conda install -c conda-forge hiplot

If you have a jupyter notebook, you can get started with something as simple as:

import hiplot as hip
data = [{'dropout':0.1, 'lr': 0.001, 'loss': 10.0, 'optimizer': 'SGD'},
        {'dropout':0.15, 'lr': 0.01, 'loss': 3.5, 'optimizer': 'Adam'},
        {'dropout':0.3, 'lr': 0.1, 'loss': 4.5, 'optimizer': 'Adam'}]
hip.Experiment.from_iterable(data).display()

See the live result

Links

• Blog post: https://ai.facebook.com/blog/hiplot-high-dimensional-interactive-plots-made-easy/
• Documentation: https://facebookresearch.github.io/hiplot/index.html
• Pypi package: https://pypi.org/project/hiplot/
• Conda package: https://anaconda.org/conda-forge/hiplot
• NPM package: https://www.npmjs.com/package/hiplot
• Examples: https://github.com/facebookresearch/hiplot/tree/main/examples

Citing

@misc{hiplot,
    author = {Haziza, D. and Rapin, J. and Synnaeve, G.},
    title = {{Hiplot, interactive high-dimensionality plots}},
    year = {2020},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\url{https://github.com/facebookresearch/hiplot}},
}

Credits

Inspired by and based on code from Kai Chang, Mike Bostock and Jason Davies.

External contributors (please add your name when you submit your first pull request):

• louismartin
• GoldenCorgi
• callistachang

License

HiPlot is MIT licensed, as found in the LICENSE file.

hiplot's Issues

Export data

Thanks for this amazing tool.
I am running it as a server but I am not able to export as expected the output of my filters.
When i click export it is opening me a new tab with the csv txt but i am not able to save this as a csv.
The saved file looks like

Am i doing anything wrong?
Would be a lot of time saved i there was an other way than copy paste
Thanks

Cannot select booleans

Whenever I use boolean as values for a parameter, I cannot select it with a range:

TypeError: Cannot read property 'from_iterable' of undefined

Issue

I made a minimalistic React based project that tries to embed Hiplot.

My repo

I followed the instructions in the documentation: NPM library (javascript).

What happened?
Installation worked

$ npm install

Starting npm worked

$ npm start

There was an error in my browser

What did you expect to happen?
I expected to see a website that had Hiplot which would be like in my notebook or like in the pip's Hiplot webserver or in like your demo.

Goal

I want a website that visualizes my data that is in CSV format with cool looking parallel plots.
I achieved this by using an <iframe /> and a separate Hiplot webserver.
But the documentation claims that your React component could also work.

Thanks

The Hiplot looks amazing and I really appreciate that there is someone maintaining it. Thank you for reading my issue.

HiPlot not displaying // Version 0.1.8.post2 broken on notebooks

When running below command:

import hiplot as hip
data = [{'dropout':0.1, 'lr': 0.001, 'loss': 10.0, 'optimizer': 'SGD'},
        {'dropout':0.15, 'lr': 0.01, 'loss': 3.5, 'optimizer': 'Adam'}]
hip.Experiment.from_iterable(data).display()

HiPlot graph never loads with following output

Loading HiPlot...
<hiplot.ipython.IPythonExperimentDisplayed at 0x1df01706b70>

Reproduction

(new_interpreter) C:\Users\erick>pip install hiplot
Collecting hiplot
  Using cached hiplot-0.1.8.post2-py3-none-any.whl (335 kB)
Requirement already satisfied: beautifulsoup4 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from hiplot) (4.9.0)
Requirement already satisfied: flask-compress in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from hiplot) (1.4.0)
Requirement already satisfied: flask in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from hiplot) (1.1.2)
Requirement already satisfied: ipython>=7.0.1 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from hiplot) (7.13.0)
Requirement already satisfied: soupsieve>1.2 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from beautifulsoup4->hiplot) (2.0)
Requirement already satisfied: click>=5.1 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from flask->hiplot) (7.1.1)
Requirement already satisfied: Werkzeug>=0.15 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from flask->hiplot) (0.16.0)
Requirement already satisfied: Jinja2>=2.10.1 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from flask->hiplot) (2.11.2)
Requirement already satisfied: itsdangerous>=0.24 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from flask->hiplot) (1.1.0)
Requirement already satisfied: decorator in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (4.4.2)
Requirement already satisfied: prompt-toolkit!=3.0.0,!=3.0.1,<3.1.0,>=2.0.0 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (3.0.5)
Requirement already satisfied: jedi>=0.10 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (0.17.0)
Requirement already satisfied: pygments in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (2.6.1)
Requirement already satisfied: colorama; sys_platform == "win32" in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (0.4.3)
Requirement already satisfied: setuptools>=18.5 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (46.1.3.post20200325)
Requirement already satisfied: traitlets>=4.2 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (4.3.3)
Requirement already satisfied: pickleshare in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (0.7.5)
Requirement already satisfied: backcall in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from ipython>=7.0.1->hiplot) (0.1.0)
Requirement already satisfied: MarkupSafe>=0.23 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from Jinja2>=2.10.1->flask->hiplot) (1.1.1)
Requirement already satisfied: wcwidth in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from prompt-toolkit!=3.0.0,!=3.0.1,<3.1.0,>=2.0.0->ipython>=7.0.1->hiplot) (0.1.9)
Requirement already satisfied: parso>=0.7.0 in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from jedi>=0.10->ipython>=7.0.1->hiplot) (0.7.0)
Requirement already satisfied: ipython-genutils in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from traitlets>=4.2->ipython>=7.0.1->hiplot) (0.2.0)
Requirement already satisfied: six in c:\users\erick\miniconda3\envs\new_interpreter\lib\site-packages (from traitlets>=4.2->ipython>=7.0.1->hiplot) (1.14.0)
Installing collected packages: hiplot
Successfully installed hiplot-0.1.8.post2

date object in parallel plot

Would it possible to handle date object ?

Is it possible to sort the table by a column by default?

I would like to have a certain column in the table sorted in descending order by default without the user clicking on it. Is this possible?

[Windows] UnicodeDecodeError

Hello, i want to try the demo but i got this issue, i try to change render.py with encoding but it not works.

i use

• Python 3.7.4
• conda 4.7.12

`import hiplot as hip

data = [{'dropout':0.1, 'lr': 0.001, 'loss': 10.0, 'optimizer': 'SGD'},
{'dropout':0.15, 'lr': 0.01, 'loss': 3.5, 'optimizer': 'Adam'},
{'dropout':0.3, 'lr': 0.1, 'loss': 4.5, 'optimizer': 'Adam'}]
hip.Experiment.from_iterable(data).display()`

UnicodeDecodeError: 'charmap' codec can't decode byte 0x9e in position 122350: character maps to <undefined>

module 'hiplot' has no attribute 'Experiment'

Hi,

I ran

• pip install hiplot
• conda install jupyter

And got ran into this error:

Any suggestions...?

Can't access off-screen columns in table when force_full_width=True

When there are too many columns to display on the screen, there's no way to access the missing ones in the table. It would be helpful if displayed columns could be toggled off or resized, or if reordering the plot columns reordered the table.

streamlit hiplot continuously rerunning

When using hiplot in streamlit with some random data, the data is regenerated and replotted every ~1s.

import numpy as np
import pandas as pd
import hiplot as hip

data = pd.DataFrame(columns=[f"x{i}" for i in range(5)], data=np.random.rand(10, 5))
xp = hip.Experiment.from_dataframe(data)
ret_val = xp.display_st(ret="selected_uids", key="hip")

I'm using streamlit version 0.64, and hiplot version 0.1.18.

Multiple parents for same data point in XY graph

The XY graph for capturing the relationship with parents and child data points are really helpful. But most of the time, we create a new child node by combining two (or more) parent nodes. In the current version, I can see

Multiple points can share the same parent.

Is it possible to include multiple parents to the same child so that the "family tree" can be traced during the experiment?

TODOLIST before release

• Make sure we don't run the server in DEBUG mode by default
• Make sure we add a warning that the server is listening to 127.0.0.1 by default
• pip install hiplot in the readme
• python -m hiplot -> hiplot
• PiPy release
• npm release
• resize stuck: on ESC stop resizing
• on refresh, keep graph sizes (even if changed run)
• on resize, dots for highlighted runs
• doc about server
• plugins (later)
• ML experiment demo
• squash commits
• /checkpoint/antares/datasets/librispeech/lists/
• selected/total
• remove dataset rendered
• pre-commit hooks, blakc etc..

QA

• Double check Jupyter notebook + force_full_width
• Double check website
• Double check Jupyter lab
• Final push to npm / pypi

Fixing axis range

I like the interactivity of the hiplot; it makes data exploration fun :)

I was wondering if it is possible to fix the axis range for (some of) numerical columns in the parallel plot. The documentation is a bit scarce on that topic.

Thanks in advance!

Buttons appear smaller when embeded in notebooks

due to Jupyter CSS

Bools are converted to NaN

True and False are both converted to NaN in the inline display and in a saved html plot in version 0.1.7.post3

import hiplot as hip
print(hip.__version__)
l = [{'a': 0, 'b':'hey', 'c':False},
     {'a': 0, 'b':'hey2', 'c':True},
]
exp_vis = hip.Experiment.from_iterable(l)
html = exp_vis.to_html('/tmp/hiplot.html')
exp_vis.display()

edited: added version

Change color map used for coloring

Hi,
Thanks for this very powerful library.

In my use case, the best and worst experiments are not very easy to distinguish with the default color map. Indeed, the best (very dark blue) and the worst (very dark red) look similar.
I am thus looking for a way to use a custom colormap (such as one available in matplotlib https://matplotlib.org/tutorials/colors/colormaps.html) that would suit better my use case.

I did not find any easy way to do so. It feels that using a custom ValueDef might be a way to go. However, it does not look the right way to go, especially since the color I want to use for coloring of continuous type and ValueDef asks for a dictionary that should provide a color for each possible value.

Is there anyway to specify the color map to be used? Otherwise, is it possible to assign manually a color to each experiment (and in such a case manually assign the right one from the colormap). I have tried adding 'c': 'black' as in https://facebookresearch.github.io/hiplot/py_reference.html#hiplot-experiment but it does not work.

Thanks

NaN values don't show in categorical column

Hi,

I've come across a weird bug with the parallel plot:
One of my float columns/axis contains NaN-values.
When plotted with Hiplot only sometimes the entry "nan/inf/null" appears on the axis; more exactly it only appears if there are at least 6 unique values other than NaN in that column.
For example, if all entries in that column contain only the values [nan 3. 5. 7. 9. 15. 30. ] the "nan/inf/null" entry shows correctly.
If however I replace all 30's with 15's without changing anything else, the entry doesn't show up.

Is that a bug, or am I overseeing something on my side?

[hiplot 0.1.19] streamlit component demo

Hi all,
I tried to run the following Hiplot's demo.

But, the demo to show Hiplot glaph in Streamlit does not work by the error shown below.

It doesn't seem to support encoding with files containing Japanese(CP932(Shift-JIS)).
I know that can solve this problem by checking "Beta: Use Unicode UTF-8 for worldwide language support" at Region, Windows Control panel.
But, this action causes another problem that other my applications will not work.
So I am looking for other ways.

Streamlit shows up, but the Hiplot graph is not displayed as shown below.

Please tell me more if you know about it.
I could really benefit from your expertise.

Font size in output

Dear,

This library is proving very useful for visualisations in my master's thesis on hyperparameter optimisation. I was wondering if there would be a way to be able to increase the font sizes of the axis labels, such that they are more readable when I include screenshots of them in my report.

Kind regards

Integration possible with Streamlit?

Hi all,
This is a really awesome project! Thanks a lot for open-sourcing it.

Given the Jupiter notebook integration, I was wondering if integration with streamlit with possible.

You may know streamlit which is kinda like a Jupyter notebook but with advantages like full python compatibility, a clean hidden state (script always run in order and smart caching) and an easy transition to production (see for instance this introductory blog post).

It would be nice to be able to use the interactive plots of HiPlot in a streamlit script, do you think this is possible?

The React module crashes in a dev server

Hi there, thanks for open-sourcing this amazing project firstly.

I tried out the react npm package, however the page crashes in browser with error like this:

×
TypeError: Cannot read property 'call' of undefined
__webpack_require__
E:/tmp/hiplot/my-app/node_modules/hiplot/dist/hiplot.lib.js:36
  33 | /******/ 		};
  34 | /******/
  35 | /******/ 		// Execute the module function
> 36 | /******/ 		modules[moduleId].call(module.exports, module, module.exports, __webpack_require__);
  37 | /******/
  38 | /******/ 		// Flag the module as loaded
  39 | /******/ 		module.l = true;
View compiled
(anonymous function)
E:/tmp/hiplot/my-app/node_modules/hiplot/dist/hiplot.lib.js:43836
  43833 | })(
  43834 |   this,
  43835 |    true && module,    // present in node.js
> 43836 |   (typeof define) == 'function' && define   // present with an AMD loader
  43837 | );
  43838 | 
  43839 | 
View compiled
(anonymous function)
E:/tmp/hiplot/my-app/node_modules/hiplot/dist/hiplot.lib.js:43841
  43838 | 
  43839 | 
  43840 | 
> 43841 | /* WEBPACK VAR INJECTION */}.call(this, __webpack_require__(5)(module)))
  43842 | 
  43843 | /***/ }),
  43844 | /* 19 */
View compiled
__webpack_require__
E:/tmp/hiplot/my-app/node_modules/hiplot/dist/hiplot.lib.js:36
  33 | /******/ 		};
  34 | /******/
  35 | /******/ 		// Execute the module function
> 36 | /******/ 		modules[moduleId].call(module.exports, module, module.exports, __webpack_require__);
  37 | /******/
  38 | /******/ 		// Flag the module as loaded
  39 | /******/ 		module.l = true;

Note that the error only appears in a development server, a production mode build works fine.

How to reproduce

The setup just follows the official react template:

npx create-react-app my-app
cd my-app
npm install hiplot -S
# add the basic example to App.js ...
npm start

I searched around and none of the answers works (on "my-app" level), maybe we need some changes in webpack building of hiplot?

Thanks!

[hiplot 0.1.17] streamlit component demo

Tried to follow instructions here: https://facebookresearch.github.io/hiplot/tuto_streamlit.html#tutostreamlit
but keep hitting

when I downloaded the bundle mentioned in the link, and include it, nothing changes. am I putting it in the wrong spot? Or is it referring to a different bundle?

Dockerfile:

FROM python:3.8-slim
RUN pip install -U streamlit
RUN pip install -U hiplot

# lines below have all been experimental to try to debug.
#USER root
# streamlit-specific commands
RUN mkdir -p $HOME/.streamlit
RUN bash -c 'echo -e "\
        [general]\n\
        email = \"\"\n\
        " > $HOME/.streamlit/credentials.toml'
RUN bash -c 'echo -e "\
        [server]\n\
        enableCORS = false\n\
        " > $HOME/.streamlit/config.toml'

COPY hiplot.bundle.js /usr/local/lib/python3.8/site-packages/hiplot/static/built/hiplot.bundle.js

Problem in using hiplot with Angular

Hi,

I tried to use hiplot with an Angular project. I did: npm install hiplot

But when I try to import hiplot through:

import * as hipl from 'hiplot';

I have problems, like the following:

../node_modules/hiplot/dist/distribution/plugin.d.ts:30:23 - error TS2503: Cannot find namespace 'JSX'.

30 render(): any[] | JSX.Element;
~~~
../node_modules/hiplot/dist/parallel/parallel.d.ts:65:15 - error TS2503: Cannot find namespace 'JSX'.

65 render(): JSX.Element;
~~~
../node_modules/hiplot/dist/plotxy.d.ts:3:15 - error TS2307: Cannot find module 'underscore'.

3 import _ from "underscore";
~~~~~~~~~~~~
../node_modules/hiplot/dist/plotxy.d.ts:53:23 - error TS2503: Cannot find namespace 'JSX'.

53 render(): any[] | JSX.Element;

How could I solve? Thanks and have a nice week

Displaying "a lot of" columns in HiPlot.

Hi,

I have a datatset with 80 columns. Is there any way to configure HiPlot such that I get a horizontal scrollbar in order to go through all columns. Right now, the depiction tries to show all columns within the screensize, which makes the app unusable.

Thank you,
Martin

hiplot command not working on windows

Hi,

I installed hiplot on windows machine using poetry. I am not able to run the hiplot from command line in my python virtual environment.

C:\Users\sarat.chinni\Codes_sequencing\hiplot>hiplot
'hiplot' is not recognized as an internal or external command,
operable program or batch file.

I have added .py extension to hiplot function in my virtual environment Scripts folder then I got the following error:

C:\Users\sarat.chinni\Codes_sequencing\hiplot>hiplot
Traceback (most recent call last):
  File "C:\Users\sarat.chinni\Codes_sequencing\biobench\sandbox\Sarat\supervised_sequencing\.venv\Scripts\hiplot.py", line 6, in <module>
    sys.exit(hip.run_server_main())
AttributeError: module 'hiplot' has no attribute 'run_server_main'

How can I solve this issue on windows? (I am able to install and hiplot in my linux machine and it worked properly)

Thank you

Allow more than 80 values for categorical variables

Hi @danthe3rd,

Thank you for the great tool for making parallel plots! 🎉

Background
I have used it for a while, and discovered that if I have more than 80 values in some categorical variable, then it is not displayed on the chart (it is still in the table).

Solution
So my suggestion would be to make this magic number 80 a parameter for display function (with default value of 80). In this way I can control chart behavior more.

Thanks!

Changing the column for distribution plot does not update the label

Unable to start webserver as per docs

I like parallel coordinate plots and was interested to try out hiplot.

The docs state starting the webserver as:

>>> hiplot

i.e. from a python prompt. When I try it just outputs:
<module 'hiplot' from 'C:\\Users\\ajohnstone\\AppData\\Local\\Continuum\\anaconda3\\envs\\visuals\\lib\\site-packages\\hiplot\\__init__.py'> and there's no server running.

I see there's a function run_server_main(), and so running hiplot.run_server_main() from the python prompt then starts the web server.

CSS is off in Safari

Parallel Plot Excludes Axes with Constant Data Points

When creating a parallel coordinates plot, if the value between data points is constant, Hiplot seems to exclude it from the graph entirely (but it is still shown in the table). Is there a way to force the axis to appear on the parallel plot? I see in the documentation that there is a way to hide certain features, but doesn't appear to be any "show" parameter.

ascending sort - uid

Hello. I just want to ask you if I can change the default sort order for uid in HiPlot. I saw that when I visualise my exp as a default there is a sorting into uid and into Dataframe that I am insert into Hiplot. I don't want to make changes by clicking the uid after plot come up. I just want to make changes that by default the plot starts with ascending order. Is this possible?
Thank you. :)

The plots are not interactive

According to this blog post I was expecting the plots to be interactive (something like bqplot or plotly with annotations and highlighting the curve when the cursor touch them)
but in my jupyter notebook the plots are static and there is no interactivity.

How to add the interactivity?

Only Export button is highlighted, Restore, Keep, and Exclude button are dim and they can not be clicked.

Installation

pip install hiplot

Code

import hiplot as hip
data = [{'dropout':0.1, 'lr': 0.001, 'loss': 10.0, 'optimizer': 'SGD'},
        {'dropout':0.15, 'lr': 0.01, 'loss': 3.5, 'optimizer': 'Adam'},
        {'dropout':0.3, 'lr': 0.1, 'loss': 4.5, 'optimizer': 'Adam'}]
hip.Experiment.from_iterable(data).display(force_full_width=True)

output

Comparison to pandas

df = pd.DataFrame(data)
pd.plotting.parallel_coordinates(
        df, 'optimizer',
        color=('#556270', '#4ECDC4', '#C7F464'))
plt.show()

pandas output

comparison

Even though hiplot is not interactive for me, still its preferable to simple method of pandas parallel plot. HiPlot has better range of values displayed for each coordinates.

comparison to plotly

import plotly.express as px

data = [{'dropout':0.1, 'lr': 0.001, 'loss': 10.0, 'optimizer': 'SGD'},
        {'dropout':0.15, 'lr': 0.01, 'loss': 3.5, 'optimizer': 'Adam'},
        {'dropout':0.3, 'lr': 0.1, 'loss': 4.5, 'optimizer': 'Adam'}]

df = pd.DataFrame(data)

fig = px.parallel_coordinates(df, 
                              color=df['optimizer'].astype('category').cat.codes,
                             color_continuous_scale=px.colors.diverging.Tealrose,
                             color_continuous_midpoint=2)
fig.show()

plotly output

Dark theme in Jupyter Lab

HiPlot's theme is broken with JupyterLab's dark mode

Specifying which field to color by

It it possible to set which field determines the colors of the lines? The first few times I used this experiments with high accuracy were read and ones with low accuracy were blue or green. This made finding useful trends stand out a bit better, but now the values of some parameter I don't care about as much determine the color and it's not as useful. Is the field chosen randomly?

2D histogram as an alternative to the scatter plot?

Are there any plans for building support for plotting 2D histograms (e.g., like the pyplot 2dhist or hexbin)? When working on larger datasets, it's difficult to estimate density in scatterplots, and adding jitter often does not help.

Exporting as a standalone html

Hi there,
thanks for the super awesome tool! I am deeply impressed by your work!

I saw some notes, that there is a way to export a plot as an interactive standalone html code.
Could you elaborate on it a little bit?

Thanks a lot!!

Allow users to hide 'uid' and 'from_uid' from the table

Hi @danthe3rd,

Thanks for the nice library. Very lightweight, very useful 🙂

Background
When I have a lot of axes on the parallel plot I always drop uid and from_uid axes, because they add some unnecessary clutter to the chart. I still want to keep these two axes in the table.
In my case I have other axes than encode runs.

Request
Add an option (maybe to the display function) to not display uid and from_uid axes on the chart.

Best,
Kamil

AttributeError: module 'hiplot' has no attribute 'Experiment'

AttributeError: module 'hiplot' has no attribute 'Experiment' with version version = "0.1.7.post2". is there any issue

HiPlot citation

Hi,

is there a preferred citation for HiPlot,
or should I just cite the blog post?

Thanks,
Mihailo

Integratable hiplot results as a React component?

Hi,
I would like to use the awesome hiplot results for a React-based blog (Gatsby).
However, integration of the fully standalone HTML page requires iframes (my current stand of knowledge).

So it would be great to have a React component, which can take the JSON generated by the python hiplot module.

I guess, you are using react components inside the framework anyway. But are there also plans to extract them for 3rd party integration?

Thanks a lot!

requirements/{dev,main}.txt are not included in sdist

setup.py reads requirements/{dev,main}.txt for install_requires and extras_require. But these two files are not included in the 1.0.3 sdist.

It's now possible to install via conda. Update README

I created a conda-forge recipe and my PR is merged so you can install this package with conda link using

conda install -c conda-forge hiplot

You might consider to update the README accordingly.

bs4 dependencies break conda recipe

I tried to build a conda-forge recipe for this package but i got this error for bs4

conda_build.exceptions.DependencyNeedsBuildingError: Unsatisfiable dependencies for platform linux-64: {'bs4'}

Is it eventyally possible to change this to beautifulsoup4? As bs4 it's jus a dummy package see this.

Saving Plot

How can I save the plot being rendered in my notebook. 'Export' only saves the data.

Use search result as filter

Hey, is there a way to use search result as a filter for the parallel plots?

Thanks

Keeping unselected lines as a background

Hello,

First of all, thank you for sharing the nice visualization library.
I really enjoy using it and I have a suggestion to improve its usability.
How about keeping unselected lines as a background when I select certain area of a variable?
In the current version, if I select certain area of a variable, Hiplot shows only selected lines.
It would be more useful if I could compare the selected ones against unselected ones.

For example, plot.ly's way as follows.

Get selection ranges

Is there a way to query the selected ranges from the parallel coordinates plots?

And thanks publishing HiPlot! Having a parallel coordinates plot for selecting data as a drop-in in ipython notebooks is really usefull.

Problem in using hiplot for streamlit

Hi, I tried to run hiplot/examples/demo_streamlit.py on Windows , I got the following error in web page:

Your app is having trouble loading the hiplot.experiment.hiplot component.

(The app is attempting to load the component from , and hasn't received its "streamlit:componentReady" message.)

If this is a development build, have you started the dev server?
If this is a release build, have you compiled the frontend?
For more troubleshooting help, please see the or visit our ***.

Is this induced by a bug? If not, how can I solve the issue?

Thank you!

HiPlot not displaying // Version 0.1.14-py3-none-any broken on jupyter notebook

Here is my setup:

OS: Windows 10
browser: Google Chrome 83.0.4103.116 (64 位）
the followings are the errors:

scroll for parallel plot

First really like Hiplot. I have used parallel plots before but really love this implementation speed and default visualization settings.
Only thing I can’t figure out is how to get the parallel plot to have a horizontal scroll bar like the info table below.