Plotly violin plots too narrow

Question

I am trying to build multiple split violin plots based on the plotly documentation example. I am not sure why the violins are so horizontally smushed. I included violinmode='overlay', which is what one other person once suggested, but it doesn't make a difference.

Below is the code I am using

lengthlm = go.Figure()

lengthlm.add_trace(go.Violin(x=lastmonth['agegroup'][lastmonth['type']=='Canine'],
                        y=lastmonth['lengthyr'][lastmonth['type']=='Canine'],
                        legendgroup='Canine', 
                        scalegroup='Canine',
                        name='Canine',
                        side='negative',
                        line_color=color_list[1]))

lengthlm.add_trace(go.Violin(x=lastmonth['agegroup'][lastmonth['type']=='Feline'],
                        y=lastmonth['lengthyr'][lastmonth['type']=='Feline'],
                        legendgroup='Feline', 
                        scalegroup='Feline',
                        name='Feline',
                        side='positive',
                        line_color=color_list[0]))

lengthlm.update_xaxes(type='category', categoryorder='array', categoryarray=['< 1 Yr','< 5 Yrs',
'< 10 Yrs','< 15 Yrs','15+ Yrs'])

lengthlm.update_traces(meanline_visible=True)
lengthlm.update_layout(violinmode='overlay',
                       violingap=0,
                      template=dash_template,
                      height=400,
                      width=750,
                      margin=dict(l=70))

lengthlm.show()

Answer 1

Within your go.Violin calls, have you explicitly tried setting width, which specifies the widths of the violins in data coordinate units?

Alternatively, you can attempt using scalemode='width' which will force all widths within a scalegroup to be the same.

See the docs.

Answer 2

• have simulated what I can imply is the structure of your dataframe
• this does not create skinny violin plots using either px or go
• hence I can only conclude that it's your data that is skinny

import pandas as pd
import numpy as np
import plotly.graph_objects as go
import plotly.express as px

# simulate data...
lastmonth = pd.DataFrame(
    {
        "agegroup": np.random.choice(
            ["< 1 Yr", "< 5 Yrs", "< 10 Yrs", "< 15 Yrs", "15+ Yrs"], 1000
        ),
        "type": np.random.choice(["Canine", "Feline"], 1000),
    }
)
lastmonth = (
    lastmonth.groupby(["agegroup", "type"])
    .apply(lambda d: d.assign(lengthyr=np.random.uniform(-1, 6, len(d))))
    .reset_index(drop=True)
)
lastmonth["agegroup"] = pd.Categorical(
    lastmonth["agegroup"],
    categories=["< 1 Yr", "< 5 Yrs", "< 10 Yrs", "< 15 Yrs", "15+ Yrs"],
    ordered=True,
)
lastmonth = lastmonth.sort_values("agegroup")

fig = (
    px.violin(lastmonth, x="agegroup", y="lengthyr", color="type")
    .update_layout(violinmode="overlay", violingap=0)
    .for_each_trace(
        lambda t: t.update(
            side="negative" if t.name == "Feline" else "positive", meanline_visible=True
        )
    )
)

fig