Forum Discussion

Anonymous's avatar
Anonymous
Not applicable
7 years ago
Solved

Use Plotly with Python Script in Power BI

Hello,

 

I'm a deep python user so would like to use it in Power BI. I have tried matplotlib and seaborn library - there are no error to use them as long as I put "plt.show()" in the scripts. However, these are all static charts and I would like to use some interactive features using Plotly Module. When I use plotly, there always shows an error of "no image created".

 

I know there is solutions in R to use Plotly but is there a solution to use Plotly with Python in Power BI?

 

Thank you.

Need Help
  • v-piga-msft's avatar
    v-piga-msft
    7 years ago

    Hi Anonymous ,

    I'm afraid that Plotly is not supported in Power BI currenly.

    Based on the blog, only the following Python packages (non-Intel MKL) are currently supported for use in your Power BI reports.

    • matplotlib
    • numpy
    • pandas
    • scikit-learn
    • scipy
    • seaborn
    • statsmodels

    In addition, you could refer to this similar thread which should make sense.

    Best Regards,

    Cherry

     

17 Replies

  • Anonymous's avatar
    Anonymous
    Not applicable
    4 years ago

    Hi any updates on how to use plotly with Python script in Power BI, I can use it but the visualization come into a browser not into the power bi dashboard.

      • Anonymous's avatar
        Anonymous
        Not applicable
        4 years ago

        Nop the only way is to do it In R not in python 

  • vishnuvvardhan's avatar
    vishnuvvardhan
    Regular Visitor
    2 years ago

    Hi  there!

     

    I recently tried to create a gauge chart with dial in Power BI using python script. There are 2 variants which I attempted:

    1. Using Matplotlib pyplot

    2. Using plotly

     

    The one using Matplotlib is actually a polar chart with it's bottom half invisible. To develop a gauge chart is more of a workaround but it's implementation in Power BI though is pretty straightforward and more compatible in terms of display. No workarounds needed! Simply create the custom visual you need and show it in the BI report.

     

    On the other hand, the variant using Plotly has better & clear features as there is a builtin functionality meant for Gauge charts. While it's super simple to create the kind of gauge chart you need, displaying it inside the python visual in PBI is a hassle! You are constantly bombarded with the following error:

     

    To solve this, I had to convert the plot to an image using kaleido.

    fig.write_image('gauge_chart.png', engine='kaleido')

     

    Then I loaded the image into Matplotlib with this:

    # Read the saved image and display it using matplotlib
img = Image.open('gauge_chart.png')
plt.imshow(img)
plt.axis('off')  # Hide the axes
plt.show()

     

    And voila! The plot will now appear in the python visual.

     

    Though the flexibility to develop better gauge charts is present in Plotly, it's implementation is quite a puzzle! And this needs more dependencies unlike the Matplotlib version!

    Had to use all of these:

    import plotly.graph_objects as go
import pandas as pd
import plotly.io as pio
from PIL import Image
import matplotlib.pyplot as plt

     

     

    Hope this helps someone!

     

    Thanks!

  • v-piga-msft's avatar
    v-piga-msft
    Resident Rockstar
    7 years ago

    Hi Anonymous ,

    I'm afraid that Plotly is not supported in Power BI currenly.

    Based on the blog, only the following Python packages (non-Intel MKL) are currently supported for use in your Power BI reports.

    • matplotlib
    • numpy
    • pandas
    • scikit-learn
    • scipy
    • seaborn
    • statsmodels

    In addition, you could refer to this similar thread which should make sense.

    Best Regards,

    Cherry

     

  • guilhermesilva's avatar
    guilhermesilva
    Regular Visitor
    4 years ago

    I am able now to make a plotly graph inside Power BI using python.
    However, Power BI shows my graph in a new window of my browser (instead of showing into the Power BI framework).

    The last line of my code is

    fig.show()

    How could I print it into the Power BI framework? 

    • HamidBee's avatar
      HamidBee
      Power Participant
      4 years ago

      I believe that this was always the case and that Plotly is still not integrated in Power BI via Python. Meaning, it still won't show graphs in Power BI. I still can't understand why Plotly works with R and not Python. To the Power BI team at Microsoft, please fix this.

    • Anonymous's avatar
      Anonymous
      Not applicable
      4 years ago

      use fig.write_image('test.png'). but it's a static image, can't be interactive.

      • HamidBee's avatar
        HamidBee
        Power Participant
        4 years ago

        I am using this example sample code:

         

        import plotly.express as px
        data_canada = px.data.gapminder().query("country == 'Canada'")
        fig = px.bar(data_canada, x='year', y='pop')
        fig.write_image('test.png')
         
        however it doesn't seem to work. When I use fig.show() it will show the chart but in a web browser.
  • stevedep's avatar
    stevedep
    Memorable Member
    4 years ago

    Hi,

     

    You could take a look at this blog post. It describes how to add an interactive Plotly Chart in Power BI. Its quite easy when you follow these steps. 

     

    Kind regards, 

     

    Steve. 

    • Anonymous's avatar
      Anonymous
      Not applicable
      4 years ago

      Steve, the link can be opened, can you please check if the link is still valid?

      Thanks,

      Cherie

      • stevedep's avatar
        stevedep
        Memorable Member
        4 years ago

        Hi,

        The link is working for me, but I also added it on the blog over here.

        Kind regards, Steve

  • _robind's avatar
    _robind
    New Member
    3 years ago

    Hi v-piga-msft, although not ideal, a static plotly image would be ok for now, but this requires fig.write_image('test.png') which requires kaleido.  What are the chances of having kaleido added to the list of approved libraries, so that users can at least view advanced plotly graphs that are not available in matplotlib or seaborn, like their go sankey plot...?