Stacked Bar Chart Alternative

segmented_total_bar_chart

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd


def segmented_total_bar_chart(
        df: pd.DataFrame,
        group_col: str,
        sub_group_col: str,
        value_col: str,
        left_margin=0.03,
        sub_group_width=0.25,
        distance=0.05,
        right_margin=0.03,
        data_label_offset=1000,
        figsize=(10, 6),
        title="Grouped Data by Sub-Group",
        title_pad=40,
        title_loc='left',
        title_x=0.04,
        legend_offset=0.1
):
    """
    Create a segmented total bar chart with dynamic spacing and labels. It is better alternative to the stacked bar chart.
    This chart allows for better readability and understanding of the data distribution across different segments.
    The chart displays total values for each group and segments within each group, with dynamic spacing and labels.
    The chart is customizable with various parameters for margins, widths, colors, and labels.
    Credit Reference: https://www.flerlagetwins.com/2025/04/the-best-alternative-to-stacked-bar.html
    :param df: DataFrame containing the data to be plotted
    :param group_col: Column name for the main grouping variable
    :param sub_group_col: Column name for the sub-grouping variable
    :param value_col: Column name for the values to be plotted
    :param left_margin: each sub-group's left margin
    :param sub_group_width: width of each sub-group
    :param distance: distance between two sub-groups in the same group
    :param right_margin: each sub-group's right margin
    :param data_label_offset: offset for data labels above the bars
    :param figsize: size of the figure
    :param title: title of the chart
    :param title_pad: padding for the title
    :param title_loc: location of the title ('left', 'center', 'right')
    :param title_x: x-coordinate for the title position
    :param legend_offset: offset for the legend text
    :return: None
    """

    # Compute total values per group and sort
    total_values = df.groupby(group_col)[value_col].sum().sort_values(ascending=False)

    # Pivot DataFrame to get sub-group values by group and sort
    sub_group_values = df.pivot(index=group_col, columns=sub_group_col, values=value_col).fillna(0).loc[
        total_values.index]

    # Define spacing parameters
    n_sub_groups = len(sub_group_values.columns)  # Number of sub-group types
    white_bar_width = (n_sub_groups - 1) * distance + n_sub_groups * sub_group_width + left_margin + right_margin

    # Plot
    fig, ax = plt.subplots(figsize=figsize)
    # Adjust subplot parameters to fit the figure. Experimental values.
    plt.subplots_adjust(left=0.01, right=0.99, top=0.875, bottom=0.04)

    # Positioning for each group
    groups = sub_group_values.index
    x_positions = np.arange(len(groups))

    # Plot total values (background gray bar) using calculated width
    ax.bar(x_positions, total_values[groups], color='lightgray', width=white_bar_width)

    # Plot sub-group bars dynamically with spacing
    colors = {sub_group: color for sub_group, color in
              zip(sub_group_values.columns, plt.cm.tab10.colors[:n_sub_groups])}
    for i, sub_group in enumerate(colors.keys()):
        offset = left_margin + i * (sub_group_width + distance) + (sub_group_width / 2) - (white_bar_width / 2)
        ax.bar(x_positions + offset, sub_group_values[sub_group], color=colors.get(sub_group, 'gray'),
               width=sub_group_width)

        # Add sub-group labels above each bar with matching color
        for x, y in zip(x_positions, sub_group_values[sub_group]):
            ax.text(x + offset, y + data_label_offset, f'${y // 1000}k', ha='center')

    # Add total labels on top (adjusted position)
    for x, y, group in zip(x_positions, total_values, groups):
        ax.text(x, y + data_label_offset, f'{group}\n${y // 1000}k', ha='center', va='bottom')

    # Labels & Formatting
    ax.set_title(title, pad=title_pad, loc=title_loc, x=title_x, color='gray', fontweight='bold', fontsize=14)
    ax.axis('off')  # Hide x-axis ticks and labels

    # Horizontal Legend with Colored Text (Top Left)
    for i, text in enumerate(colors.keys()):
        ax.text(title_x + (i * legend_offset), 1.06, text, transform=ax.transAxes, fontweight='bold',
                color=colors[text])

    plt.show()


data = {
    'Region': ['West', 'West', 'West', 'East', 'East', 'East', 'Central', 'Central', 'Central', 'South', 'South',
               'South'],
    'Segment Type': ['Consumer', 'Corporate', 'Home Office', 'Consumer', 'Corporate', 'Home Office', 'Consumer',
                     'Corporate', 'Home Office', 'Consumer', 'Corporate', 'Home Office'],
    'Sales': [364000, 232000, 143000, 357000, 204000, 131000, 254000, 158000, 91000, 196000, 122000, 74000],
}

df = pd.DataFrame(data)
segmented_total_bar_chart(
    df,
    group_col='Region',
    sub_group_col='Segment Type',
    value_col='Sales',
    title="Regions-Revenue by Segment",
)