Plotly Radar Chart: arranging y-axis labelled with string by levels

Question

Assume I have a df:

df = pd.DataFrame({'category': ['Apple', 'Pear', 'Banana', 'Orange', 'Cherry'],
                   'value': [1, 1, 3, 2, 0],
                  'mark': ['Average', 'Average', 'Terrible', 'Bad', 'Good']})

What I want to do is plot df on radar chart via Plotly. The outcome looks like this:

Except this round, instead of numbers, I'm replacing the y-axis or radius with strings. For example: 0 is 'good', 1 is 'average', 2 is 'bad', & 3 is 'terrible'. However once I plot it, the orders were randomized:

How do I re-sort the y-axis?

This is the code I'm currently using:

import plotly.express as px
display(px.line_polar(df.sort_values(by= 'category'), theta= 'category', r= 'mark').update_traces(fill='toself'))

Answer 1

Use two primary techniques

1. consider mark to be a categorical in dataframe. Use ordered index to build polar figure
2. update radialaxis from codes assigned in step 1 back to text

import pandas as pd
import plotly.express as px

df = pd.DataFrame({'category': ['Apple', 'Pear', 'Banana', 'Orange', 'Cherry'],
                   'value': [1, 1, 3, 2, 0],
                  'mark': ['Average', 'Average', 'Terrible', 'Bad', 'Good']})

# need order of categoricals...
cat = ['Good', 'Average', 'Bad', 'Terrible']

# output radial as an ordered number
df2 = df.assign(mark=pd.Categorical(df["mark"], ordered=True, categories=cat).codes).sort_values(by= 'category')
fig = px.line_polar(df2, theta= 'category', r= 'mark').update_traces(fill='toself')

# change axis back to text
fig.update_layout(polar={"radialaxis":{"tickmode":"array","tickvals":[i for i in range(len(cat))],"ticktext":cat}})

fig

without use of categorical

import pandas as pd
import plotly.express as px

df = pd.DataFrame({'category': ['Apple', 'Pear', 'Banana', 'Orange', 'Cherry'],
                   'value': [1, 1, 3, 2, 0],
                  'mark': ['Average', 'Average', 'Terrible', 'Bad', 'Good']})


# output radial as an ordered number
fig = px.line_polar(df.sort_values("category"), theta= 'category', r= 'value').update_traces(fill='toself')

# get mapping of value to mark
df2 = df.loc[:,["value","mark"]].drop_duplicates().sort_values("value")
# change axis to text
fig.update_layout(polar={"radialaxis":{"tickmode":"array","tickvals":df2["value"],"ticktext":df2["mark"]}})

fig