Windrose example from plot.ly not working

[rr5577]

I just did a fresh install to try plotly.py and wanted to try the pandas windrose example from the plotly site ( https://plot.ly/pandas/wind-rose-charts/ ).

import pandas as pd
import colorlover as cl
import numpy as np
import plotly.plotly as py
import plotly.graph_objs as go


df = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/wind_rose.csv')
df.head()

data = []
counter = 0
for col in df.columns:
    if col != 'r':
        data.append(
            go.Area(t=df['r'],
                    r=df[col],
                    marker=dict(color=cl.scales['9']['seq']['PuBu'][counter]),
                    name=col+' m/s' ) )
        counter+=1
layout = go.Layout(
    title='test wr',
    font=dict(
        size=16
    ),
    
    radialaxis=dict(
        ticksuffix='%'
    ),
    orientation=270,
    barmode="stack"
    
)
fig = go.FigureWidget(data=data, layout=layout)
fig

but i only get a blank output cell.

if i remove the barmode="stack" kwarg, the plot renders, but without data:

Plotly Version: 3.1.1

Update:
i managed to get the plot and data to display, it seems using the pandas columns for "r" and "t" values doesn´t work (as in #1116).

However, dynamically updating the plot (e.g. changing title color) doesn´t work.

[jonmmease]

Hi @rr5577 , thanks for taking the time to report this.

Yeah, it looks like the Plotly.js JavaScript library is choking when plotly.py is passed a pandas series or numpy array. And I can also reproduce you're issue of not being able to update the title properties.

These issues should be addressed when this trace type is replaced with a new trace that uses the Polar 2.0 infrastructure. See plotly/plotly.js#2810 for some discussion.

I'll leave this issue open until the new trace is released and included with plotly.py.

[jonmmease]

plotly.py 3.3 is going to include plotly.js 1.41 which includes the new barpolar trace type (plotly/plotly.js#2954). This is the much more capable successor to the polar area traces used in this example.

[rr5577]

Thanks for the update, looking forward to trying it!

[jonmmease]

barpolar trace included in upgrade to plotly.js 1.41.3 in #1202, will be released in plotly.py 3.3.0 soon.

import plotly.graph_objs as go
fig = go.FigureWidget()
fig.add_barpolar(r=[1, 3, 2], theta=[0, 45, 90])
fig.add_barpolar(r=[3, 5, 1], theta=[0, 45, 90])
fig.add_barpolar(r=[2, 1, 6], theta=[0, 45, 90])
fig

See plotly/documentation#1100 for progress on updating the windrose documentation to use barpolar. Thanks for reporting!

[rr5577]

This works great now (dynamically updating the plot as well). Only the code from the python example specifying the (layout) axis seems to be obsolete (it doesn´t do anything when changed).
After changing that it worked:

layout = go.Layout(
    title='test wd',
    font=dict(size=16), 
    polar=dict(
        radialaxis=dict(ticksuffix="%",angle=45,tickfont=dict(size=12)),
        angularaxis=dict(direction="clockwise",tickfont=dict(size=14)),
    ),