8

I'm trying to build a Power BI tool for some data analysis, and one of the plots I need is an inverse quantile plot (quantiles on x axis, values on y axis). Power BI does not have this, and I can't find one on the app marketplace, so am using Python to code up what I need.

The static plot from pandas.DataFrame.plot() works fine but lacks the pinache of an interactive plot. I've coded up the plot I need using plotly, and ran it with py.iplot(), but Power BI tells me

No image was created. The Python code didn't result in creation of any visuals. Make sure your Python script results in a plot to the Python default device

There was no error, and I confirmed the code is fine by running the plot using py.plot(), and viewed the result in the browser. My code is:

import plotly.plotly as py
import plotly.graph_objs as go

# get the quantiles and reshape
qs = dataset.groupby(by='HYDROCARBON_TYPE').Q42018_AbsDevi.quantile(q=[0.01,0.05,0.1,0.2,0.25,0.5,0.75,0.8,0.9,0.95,0.99]).unstack().transpose()
# plot it
traces = []
for col in qs.columns:
    traces.append(go.Scatter(x=qs.index, y=qs[col], name=col))
py.plot(traces,filename='basic-line')

Why would this not be working?

Dr. Andrew
  • 2,532
  • 3
  • 26
  • 42

2 Answers2

4

I wasn't able to find a solution using PowerBI, Plotly and Python, nor was I able to reproduce your errors. Regarding your errors, I ended up with visualizations that were either timed out or reporting a data type error. But we can get back to that if that's still interesting after another suggested solution, because I have been able to produce an interactive q-plot using PowerBI, plotly, ggplot and an R script visual like this:

enter image description here

Assuming that your main priorities are to make an interactive quantile plot in PowerBI, and that Python as a tool comes second, just follow the steps outlined in this post, and replace the R script with this:

source('./r_files/flatten_HTML.r')

############### Library Declarations ###############
libraryRequireInstall("ggplot2");
libraryRequireInstall("plotly")
####################################################

################### Actual code ####################
df <- data.frame(y = Values$Data)

# Build basic ggplot
g <- ggplot(df, aes(sample = y))

# Add quantile details
g = g + stat_qq() + stat_qq_line()

############# Create and save widget ###############
p = ggplotly(g);
internalSaveWidget(p, 'out.html');
####################################################

That should do the trick. Don't hesitate to let me know if this does not work for you.

vestland
  • 55,229
  • 37
  • 187
  • 305
  • This does not meet my needs, as I'm not an R developer, do not have anything R installed (and neither do my users/collaborators) and am unable to install additional javascript resources. – Dr. Andrew Mar 11 '19 at 08:53
  • @Dr. Drew Thanks for your feedback. In that case I'm hoping for a better answer just as much as you are! – vestland Mar 11 '19 at 09:18
  • @Dr. Drew Have you considered a bounty for this question? – vestland Apr 10 '19 at 06:33
  • Thanks for the suggestion, but it's not that important to me that I want to sacrifice rep. The workaround that I'm willing to live with - though it's not a solution - is to just have my plot open in the default browser. – Dr. Andrew Apr 11 '19 at 07:48
  • 1
    @Dr. Drew Here's a suggestion: I will start a bounty on your question. If you receive an answer, that's great for all of us. I take the rep loss, and thats fine. If you get no answer that is better than my suggestion, then you mark it as the accepted answer. since it would be fair to assume that there ***is*** no such option. Sound good? Oh, and if a better solution turns up after the bounty period is out, it would only be fair to unaccept my answer and award the better solution with the acceptance mark. – vestland Apr 11 '19 at 07:57
  • This was an amazing deal, not sure why it did not go ahead, A similar problem here , https://stackoverflow.com/questions/60214526/custom-visual-in-power-bi-using-python?noredirect=1#comment106506937_60214526 – Mohsen Sichani Feb 13 '20 at 20:53
  • @mohsenhs Yeah, I thought so too. I guess OP just decided to accept things as they were and go about his day. I hope you'll get a good answer to the question you just linked there though (+1). – vestland Feb 13 '20 at 20:58
0

You can take a look at this blog post. It describes how to add an interactive javascript Plotly Chart in Power BI. Its quite easy.

Kind regards,

Steve.

Steve de Peijper
  • 76
  • 1
  • 8