how to plot a single line in plotly with multiple colors according to a categorical variable

Question

How can I get a single connected line in plotly with different colors?

The plot below shows an attempt at a solution. However, the line has an ugly break between point 10 and point 90. How can I have a single line with multiple colors according to a categorical variable without breaking?

import numpy as np
import random
import pandas as pd
import plotly.express as px

white_noise = np.array([random.gauss(mu=0.0, sigma=1.0) for x in range(100)])
rw = white_noise.cumsum()
rw_df = pd.DataFrame({
    'random_walk': rw, 
    'color': 10*['black'] + 50 * ['blue'] + 40*['black'], 
    'x': range(100)
})```

fig = px.line(rw_df, x='x', y='random_walk', color='color')
fig.show()

Answer 1

See if this is what you are looking for...

First - break the dataframe into an array of dataframes, each with a set of same colored rows... in your example, it would be 3 dataframes - first for the black line, second for blue and then another for the black.

Second - Plot the first item as a line. Then for each subsequent dataframes, use add scatter to plot a new line.

import numpy as np
import random
import pandas as pd
import plotly.express as px

white_noise = np.array([random.gauss(mu=0.0, sigma=1.0) for x in range(100)])
rw = white_noise.cumsum()
rw_df = pd.DataFrame({
    'random_walk': rw, 
    'color': 10*['black'] + 50 * ['blue'] + 40*['black'], 
    'x': range(100)
})

## Break your dataframe into an array of smaller dataframes, each with single color
rw_df['group']=rw_df['color'].ne(rw_df['color'].shift()).cumsum()
rw_df
rw_df = rw_df.groupby('group')
dfs = []
for name, data in rw_df:
    dfs.append(data)

## Plot the first line - dfs[0]    
fig = px.line(dfs[0], x='x', y='random_walk')
fig.update_traces(line_color=dfs[0]['color'].iloc[0])

## For other lines in array, plot new line
for i in range(1, len(dfs)):
    fig.add_scatter(x=dfs[i]["x"], y=dfs[i]["random_walk"], 
                    line_color=dfs[i]['color'].iloc[0], 
                    name=dfs[i]['color'].iloc[0])
    
fig.show()