[FIXED] Iterate list of colors based on unique column value using a value from a second column to differ between dark and light colors - Jupyter Notebook

Issue

I am using Jupyter notebook. I have combined two DataFrames together, so in order to differentiate between the two I have created a new third column 'Color Code' where there are two values 'a' and 'b'. I have no problem iterating through the list of colors I have defined but I need to be able to color dark vs light depending on the unique columns and 'Color Code' column.

I have an MRE below. Apologies for the big example DataFrame, just need to be able to showcase what I am trying to accomplish.

df1 = pd.DataFrame({'x1':['a', 'a', 'a', 'a', 'a', 'a', 'b', 'b', 'b', 'b', 'b', 'b', 'c', 'c', 'c', 'c', 'c', 'c', 'c', 'c', 'd', 'd', 'd', 'd'],
                     'x2':['3','4','8','0','11','1','15','5','67','3','1','4','5','88','9','3','7','2','3','43','34','56','96','26'],
                     'Color Mode':['a','a','b','b','b','b','a','a','a','a','b','b','a','b','b','b','b','b','b','b','a','a','a','b']})

def Color_Unique(s):
    df = s.copy()
    
    color_map1 = {}
    color_map2 = {}

    x = pd.DataFrame({'Color Mode': ['a']})
    y = pd.DataFrame({'Color Mode': ['b']})
    
    Trade_Cusip_Combo_Key = df['x1'].unique()
    
    if x.any(axis=None):
        colors_to_use = ['background-color: #ADD8E6', 'background-color: #90ee90', 'background-color: #FFD580', 'background-color: #CBC3E3', 'background-color: #D3D3D3', 'background-color: #C4A484']
        colors_cycle = cycle(colors_to_use)

        for Trade_Cusip_Combo in Trade_Cusip_Combo_Key:
            color_map1[Trade_Cusip_Combo] = next(colors_cycle)

        for index, row in df.iterrows():
            if row['x1'] in Trade_Cusip_Combo_Key:
                Trade_Cusip_Combo = row['x1']
                my_color = color_map1[Trade_Cusip_Combo]
                df.loc[index,:] = my_color
            else:
                df.loc[index,:] = 'background-color: '        
        return df
    elif y.any(axis=None):
        colors_to_use = ['background-color: #ADD8E6', 'background-color: #90ee90', 'background-color: #FFD580', 'background-color: #CBC3E3', 'background-color: #D3D3D3', 'background-color: #C4A484']
        colors_cycle = cycle(colors_to_use)

        for Trade_Cusip_Combo in Trade_Cusip_Combo_Key:
            color_map2[Trade_Cusip_Combo] = next(colors_cycle)

        for index, row in df.iterrows():
            if row['x1'] in Trade_Cusip_Combo_Key:
                Trade_Cusip_Combo = row['x1']
                my_color = color_map2[Trade_Cusip_Combo]
                df.loc[index,:] = my_color
            else:
                df.loc[index,:] = 'background-color: '        
        return df
    else:
        print("boken")

df4 = df1.style.apply(Color_Unique, axis=None)

df4

This is what my code currently looks like when you run my above MRE.

This is what I need it to look like:

You will notice dark vs light colors alternate when column 'x1' values are the same but column 'Color Mode' values are either 'a' (dark color) or 'b' (lighter color). I need my function to be able to iterate through my defined dark and light colors based on column 'x1' and column ' Color Mode' as well as skip a line after each group is colored (blue to green to orange to purple).

Solution

Before turning to the "styling" part of the exercise, let's use a function to insert blank rows into your df. We can do this as follows:

import pandas as pd

df1 = pd.DataFrame({'x1':['a', 'a', 'a', 'a', 'a', 'a', 'b', 'b', 'b', 
                          'b', 'b', 'b', 'c', 'c', 'c', 'c', 'c', 'c', 
                          'c', 'c', 'd', 'd', 'd', 'd'],
                     'x2':['3','4','8','0','11','1','15','5','67','3',
                           '1','4','5','88','9','3','7','2','3','43',
                           '34','56','96','26'],
                     'Color Mode':['a','a','b','b','b','b','a','a','a',
                                   'a','b','b','a','b','b','b','b','b',
                                   'b','b','a','a','a','b']})

def blank_rows(df, col_name):
    insert_idx = [v[-1]+1 for v in df.groupby(col_name).groups.values()]
    for i, idx in enumerate(insert_idx[:-1]):
        temp = pd.DataFrame({k: '' for k in df.columns}, index=[idx+i])
        df = pd.concat([df.iloc[:idx+i], temp, df.iloc[idx+i:]])\
        .reset_index(drop=True)
    return df

# add `col_name` (e.g. `x1`) to specify which groups need to be split up
df = blank_rows(df1, 'x1')

# N.B. the groups inside column `col_name` need to be sorted, 
# and the index needs to be "regular" (i.e. `0,1,2...`).
# E.g. `df = blank_rows(df1, 'Color Mode')` would get you a nonsensical result;
# Instead you would want to use: 
# `df = blank_rows(df1.sort_values('Color Mode').reset_index(drop=True), 
#                'Color Mode')`

df will now contain empty rows like below:

df.iloc[[5,6,7,12,13,14],:]

Now, let's apply a function to take care of the styles. E.g.:

from itertools import cycle

def color_unique(df_input):
    df = df_input.copy()
    colors = ['#8EA9DB', '#D9E1F2', # blues
              '#A9D08E', '#E2EFDA', # greens
              '#F4B084', '#FCE4D6', # reds
              '#9751CB', '#DEC8EE'] # purples

    colors_cycle = cycle(colors)
        
    groups = df.groupby(['x1','Color Mode']).groups
    for k, v in groups.items():
        if '' in k:  
            df.iloc[v] = f'background-color: white;'
        else:
            df.iloc[v] = f'background-color: {next(colors_cycle)};'
    return df

df_styled = df.style.apply(color_unique, axis=None)
df_styled

Result:

Explanation

• Create a list with dark/light color pairs. Since we are using cycle from itertools, the list does not need to be exhaustive. (N.B.: the above function is expecting that each group in x1 is always split in 2 subgroups based on Color Mode. E.g. if you expect groups that only have a, not b, the above function would need to be adjusted!)
• Use df.groupby to isolate the different groups. Make sure to add sort=False to preserve the original order of the groups.
• The Groupby.groups property stores a dict with "group name" (e.g. ('a', 'a')) as keys and the associated index values as values (e.g. for ('a', 'a') the value is Int64Index([0, 1], dtype='int64').
• We want to use the values per group to select slices consecutively from the df with .iloc and assign a color. However, we want to skip the group that has key (''),(''), For that reason, we want to iterate over group.items() and catch both the (k)ey and (v)alue. If the key (tuple) contains an empty string, we color the row "white", else we assign the next color from our list of colors.

Answered By - ouroboros1