How to set specific color to some bars in a plotly bar graph?

Question

I'm trying to set different colors for some bars in a plotly express bar graph:

import plotly.express as px
import pandas as pd

data = {'Name':['2020/01', '2020/02', '2020/03', '2020/04',  
         '2020/05', '2020/07', '2020/08'],  
        'Value':[34,56,66,78,99,55,22]}
df = pd.DataFrame(data)

color_discrete_sequence = ['#ec7c34']*len(df)
color_discrete_sequence[5] = '#609cd4'
fig=px.bar(df,x='Name',y='Value',color_discrete_sequence=color_discrete_sequence)
fig.show()

My expectations were that one (the sixth one) bar had a different color, however I got this result:

What am I doing wrong?

Answer 1

This happens because color in px.bar is used to name a category to illustrate traits or dimensions of a dataset using a colorscale. Or in you your case, rather a color cycle since you're dealing with a categorical / discrete case. color_discrete_sequence is then used to specify which color sequence to follow. One way to achieve your goal using your setup here, is to simply define a string variable with unique values, for example like df['category'] [str(i) for i in df.index], and then use:

fig=px.bar(df,x='Name',y='Value',
           color = 'category',
           color_discrete_sequence=color_discrete_sequence,
           )

Plot:

If df['category'] is a numerical value, color_discrete_sequence will be ignored, and a default continuous sequence will be applied:

If anything else is unclear, don't hesitate to let me know.

Complete code:

import plotly.express as px
import pandas as pd

data = {'Name':['2020/01', '2020/02', '2020/03', '2020/04',  
         '2020/05', '2020/07', '2020/08'],  
        'Value':[34,56,66,78,99,55,22]}
df = pd.DataFrame(data)
df['category'] = [str(i) for i in df.index]
# df['category'] = df.index

color_discrete_sequence = ['#ec7c34']*len(df)
color_discrete_sequence[5] = '#609cd4'
fig=px.bar(df,x='Name',y='Value',
           color = 'category',
           color_discrete_sequence=color_discrete_sequence,
           )
fig.show()