HoloViews: create boxplots interactive

Question

I'd like to create a Dashboard where the user can choose which data is shown in a Boxplot via chosing by button. All I could find where instructions for linear-interactive plots, unfortunately i am not able to find out how to make interactive boxplots.

so far my code looks like this:

import numpy as np
import hvplot.pandas 
import panel as pn 
import holoviews as hv

np.random.seed(1234)
adf = pd.DataFrame(np.random.rand(3, 4),
                  columns=['Col1', 'Col2', 'Col3', 'Col4'])

bdf = pd.DataFrame(np.random.rand(3, 4),
                  columns=['Col1', 'Col2', 'Col3', 'Col4'])

adf['Type'] = 'A'
bdf['Type'] = 'B'
df = pd.concat([adf,bdf])
df.index.names = ['idx']

That's my DataFrame:

idx Col1        Col2        Col3        Col4        Type
0   0.191519    0.622109    0.437728    0.785359    A
1   0.779976    0.272593    0.276464    0.801872    A
2   0.958139    0.875933    0.357817    0.500995    A
0   0.683463    0.712702    0.370251    0.561196    B
1   0.503083    0.013768    0.772827    0.882641    B
2   0.364886    0.615396    0.075381    0.368824    B

created some buttons:


typebutton = pn.widgets.RadioButtonGroup(name='Typebutton', options=['A','B'],button_type='success')

and now connect the buttons with the dataframe:

df = df.interactive()
df_inter = (df[df.Type == typebutton])

Interactive DataFrame with buttons

now i prepare for plotting:


df_group = df_inter.groupby(['idx'])[['Col1', 'Col2', 'Col3', 'Col4']].mean()

and at this point i can use

dfplot = df_group.hvplot()

for getting an interactive linear plot of my Data. But how can I create Interactive Boxplots?

So far i know that:

hv.BoxWhisker(bdf.melt(), kdims='variable', vdims='value')

Is giving me the Boxplots, but how do i make them interactive?

score 0 · Accepted Answer · answered Jul 15 '22 at 15:29

Box plots are supported by hvPlot; no need to drop down to HoloViews. See https://hvplot.holoviz.org/reference for all the plot types supported. Note that the .hvplot.box() syntax listed on the web pages is not yet supported for .interactive; you need to call it as kind='box' instead:

import numpy as np, pandas as pd, hvplot.pandas, panel as pn
np.random.seed(1234)
cols = ['Col1', 'Col2', 'Col3', 'Col4']
adf = pd.DataFrame(np.random.rand(3, 4), columns=cols)
bdf = pd.DataFrame(np.random.rand(3, 4), columns=cols)
adf['Type'] = 'A'
bdf['Type'] = 'B'

df = pd.concat([adf,bdf])
df.index.names = ['idx']

typebutton = pn.widgets.RadioButtonGroup(name='Typebutton', options=['A','B'],button_type='success')
dfi = df.interactive(width=800)
df_inter = (dfi[dfi.Type == typebutton])
df_inter.hvplot(kind='box')

unfortunately i get : ```TypeError: '>=' not supported between instances of 'Version' and 'str'``` when i run the ```df_inter.hvplot(kind='box')``` — Paul Borowy, Jul 18 '22 at 06:35
```df_inter.hvplot(kind='box', legend=False)``` is doing the Trick!! — Paul Borowy, Jul 18 '22 at 07:33

HoloViews: create boxplots interactive

1 Answers1