Percent Stacked Bar Chart

📊 Plotly Python
1

Hi everyone,
Is there any direct way to produce a percentage stacked bar chart with Plotly Express?

I have the following dataframe

candidato	Page Name	Likes
0	AndrĂ© Ventura	AgĂȘncia Lusa	95
1	André Ventura	CM TV	2120
2	André Ventura	CNN Portugal	13825
3	André Ventura	Correio da Manhã	5579
4	André Ventura	Diårio de Notícias	1381
5	André Ventura	ECO	101
6	André Ventura	Expresso	4070
7	André Ventura	Jornal Económico	665
8	André Ventura	Jornal de Negócios	1355

The dataframe repeats for 8 more candidatos and has 22 Page Names

The result I’m trying to achieve is this

Likes by media outlet
Likes by media outlet1210×649 41.8 KB

The reason for this is that a regular stacked bar chart fails to give the dimensions between candidates, as you can see in this example.

newplot (2)
newplot (2)1568×525 55.7 KB

Is there any way to do this?

Thank you in advance!

Disclosure: This is part of a media monitor report that I’m voluntarily preparing, The final product will be posted under a creative commons license, and there is no commercial intent.

1 Like
2

Hi @jgomes_eu,

To my knowledge there isn’t a direct way to do it, but it is very straightforward to transform the data in pandas. See for instance this SO post.

3

Hi, thank @jlfsjunior

The solution presented really doesn’t work for me, as my dataset is a bit more complex than that.

Totals have to be calculated for each Page Name for each of column (Total Interactions,Likes,Shares,Love,Angry), and then percentages calculated.

Is this something that is on the roadmap for Plotly.Express?
It would be really useful as this is a type of chart that is commonly used for data visualization.

4

Assuming your dataset resides in a dataframe named data, the following commands will create the graph you are looking for:

df = data.groupby([‘Page Name’, ‘candidato’]).agg({‘Likes’: ‘count’})
df[‘prop’] = df.groupby(level=0).apply(lambda x: 100*x/x.sum()).reset_index(level=0, drop=True)
fig = px.bar(df.reset_index(), x=‘Page Name’, y=‘prop’, color=‘candidato’)
fig.show()

I hope it helps, even though it is a bit too late.

1 Like