Mark_area() color undefined with correct column name

[lborcard]

What happened?

Trying to add color by group but it shows color as undefined instead of displaying the color differences.

_chart_loss2 = (

    alt.Chart(sub_loss_real).transform_density("Percentage_loss", as_=["Percentage_loss", "density"])
    .mark_area()
    .encode(
        alt.X("Percentage_loss:Q", scale=alt.Scale(domain=[0, 1])),
        alt.Y("density:Q"),
        alt.Color("Percentage_loss_dim_2:N")
    )
)
_chart_loss2

info about the dataframe
<class 'pandas.core.frame.DataFrame'>
Index: 5000 entries, 2003 to 10552
Data columns (total 4 columns):
 #   Column                 Non-Null Count  Dtype  
---  ------                 --------------  -----  
 0   chain                  5000 non-null   int64  
 1   draw                   5000 non-null   int64  
 2   Percentage_loss_dim_2  5000 non-null   string 
 3   Percentage_loss        5000 non-null   float64
dtypes: float64(1), int64(2), string(1)
memory usage: 195.3 KB

What would you like to happen instead?

I would like the plot to show me a two different colors for each group and I would like to be able to know what is wrong with my input since I tried every way possible to modify my column so that it can separate it by group.

Which version of Altair are you using?

5.3.0

[dsmedia]

The examples here and here use the groupby parameter. Does this sample code (modeled after yours but with a sample dataset) generate the result you're looking for?

import altair as alt
import pandas as pd
import numpy as np

# Create a sample dataset
np.random.seed(42)  # for reproducibility
n = 1000
sub_loss_real = pd.DataFrame({
    'Percentage_loss': np.concatenate([
        np.random.beta(2, 5, n),  # distribution for group A
        np.random.beta(5, 2, n)   # distribution for group B
    ]),
    'Percentage_loss_dim_2': np.repeat(['Group A', 'Group B'], n)
})

# Create the chart
chart_loss = (
    alt.Chart(sub_loss_real)
    .transform_density(
        'Percentage_loss',
        groupby=['Percentage_loss_dim_2'],
        as_=['Percentage_loss', 'density']
    )
    .mark_area(opacity=0.5)
    .encode(
        x=alt.X('Percentage_loss:Q', scale=alt.Scale(domain=[0, 1])),
        y=alt.Y('density:Q'),
        color=alt.Color('Percentage_loss_dim_2:N', scale=alt.Scale(scheme='category10'))
    )
    .properties(
        width=600,
        height=400,
        title="Density Plot of Percentage Loss by Group"
    )
)

# Display the chart
chart_loss.show()

[joelostblom]

Closing a solution has been posted above.

[lborcard]

Thanks for help, I somehow found out that you needed groupby. Imo using color should by default do the grouping.

[joelostblom]

We're working including a dedicated density mark in Vega-Lite / Altair vega/vega-lite#3442. This would support grouping via color directly and without the need to use the transform at all.