Название: Как визуализировать иерархические данные с помощью вложенных круговых диаграмм в Python?

Один из способов сделать это — определить функцию, которая создает вложенную круговую диаграмму:

import pandas as pd
import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1.inset_locator import inset_axes

data = {
    'Phylum': ['Proteobacteria', 'Proteobacteria', 'Proteobacteria', 'Proteobacteria',
               'Firmicutes', 'Firmicutes', 'Actinobacteria', 'Proteobacteria',
               'Firmicutes', 'Proteobacteria'],
    'Genus': ['Pseudomonas', 'Klebsiella', 'Unclassified', 'Chromobacterium',
              'Lysinibacillus', 'Weissella', 'Corynebacterium', 'Cupriavidus',
              'Staphylococcus', 'Stenotrophomonas'],
    'Species': ['Unclassified', 'Unclassified', 'Unclassified', 'Unclassified',
                'boronitolerans', 'ghanensis', 'Unclassified', 'gilardii',
                'Unclassified', 'geniculata'],
    'Absolute Count': [73745, 10777, 4932, 1840, 1780, 1101, 703, 586, 568, 542]
}

df = pd.DataFrame(data)


def create_nested_pie(df):
    fig, ax = plt.subplots()
    size = 0.3
    phylum_counts = df.groupby('Phylum')['Absolute Count'].sum()
    phylum_labels = phylum_counts.index.tolist()
    ax.pie(phylum_counts, labels=phylum_labels, radius=1, wedgeprops=dict(width=size, edgecolor='w'))

    firmicutes_genus_counts = df[df['Phylum'] == 'Firmicutes'].groupby('Genus')['Absolute Count'].sum()
    firmicutes_genus_labels = firmicutes_genus_counts.index.tolist()
    ax.pie(firmicutes_genus_counts, labels=firmicutes_genus_labels, radius=1-size, wedgeprops=dict(width=size, edgecolor='w'),
           labeldistance=0.7)

    lysinibacillus_species_counts = df[(df['Phylum'] == 'Firmicutes') & (df['Genus'] == 'Lysinibacillus')].groupby('Species')['Absolute Count'].sum()
    lysinibacillus_species_labels = lysinibacillus_species_counts.index.tolist()
    ax.pie(lysinibacillus_species_counts, labels=lysinibacillus_species_labels, radius=1-2*size, wedgeprops=dict(width=size, edgecolor='w'),
           labeldistance=0.4)

    plt.show()

create_nested_pie(df)

Что дает вам:

Обновление: фильтрация

Если ваши данные большие, то фильтрацию для отображения определенных меток можно выполнить, слегка изменив функцию:

import pandas as pd
import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1.inset_locator import inset_axes

data = {
    'Phylum': ['Proteobacteria', 'Proteobacteria', 'Proteobacteria', 'Proteobacteria',
               'Firmicutes', 'Firmicutes', 'Actinobacteria', 'Proteobacteria',
               'Firmicutes', 'Proteobacteria'],
    'Genus': ['Pseudomonas', 'Klebsiella', 'Unclassified', 'Chromobacterium',
              'Lysinibacillus', 'Weissella', 'Corynebacterium', 'Cupriavidus',
              'Staphylococcus', 'Stenotrophomonas'],
    'Species': ['Unclassified', 'Unclassified', 'Unclassified', 'Unclassified',
                'boronitolerans', 'ghanensis', 'Unclassified', 'gilardii',
                'Unclassified', 'geniculata'],
    'Absolute Count': [73745, 10777, 4932, 1840, 1780, 1101, 703, 586, 568, 542]
}

df = pd.DataFrame(data)

def create_filtered_nested_pie(df, phylum_filter=None, genus_filter=None, species_filter=None):
    fig, ax = plt.subplots()
    size = 0.3
    
    if phylum_filter is not None:
        df = df[df['Phylum'].isin(phylum_filter)]
    phylum_counts = df.groupby('Phylum')['Absolute Count'].sum()
    ax.pie(phylum_counts, labels=phylum_counts.index.tolist(), radius=1, 
           wedgeprops=dict(width=size, edgecolor='w'))

    if genus_filter is not None:
        df_genus = df[df['Genus'].isin(genus_filter)]
    else:
        df_genus = df
    genus_counts = df_genus.groupby('Genus')['Absolute Count'].sum()
    ax.pie(genus_counts, labels=genus_counts.index.tolist(), radius=1-size, 
           wedgeprops=dict(width=size, edgecolor='w'), labeldistance=0.7)

    if species_filter is not None:
        df_species = df_genus[df_genus['Species'].isin(species_filter)]
    else:
        df_species = df_genus
    species_counts = df_species.groupby('Species')['Absolute Count'].sum()
    ax.pie(species_counts, labels=species_counts.index.tolist(), radius=1-2*size, 
           wedgeprops=dict(width=size, edgecolor='w'), labeldistance=0.4)

    plt.show()

create_filtered_nested_pie(df, 
                           phylum_filter=['Proteobacteria', 'Firmicutes'],
                           genus_filter=['Pseudomonas', 'Lysinibacillus'],
                           species_filter=['boronitolerans', 'Unclassified'])

что дает

Чтобы получить пропорциональную круговую диаграмму и исключить все, что не фильтруется (становясь затем прозрачным), я немного изменил свою функцию, чтобы вычислить занятую пропорцию и «скрыть» то, что не фильтруется:

import pandas as pd
import matplotlib.pyplot as plt

data = {
    'Phylum': ['Proteobacteria', 'Proteobacteria', 'Proteobacteria', 'Proteobacteria',
               'Firmicutes', 'Firmicutes', 'Actinobacteria', 'Proteobacteria',
               'Firmicutes', 'Proteobacteria'],
    'Genus': ['Pseudomonas', 'Klebsiella', 'Unclassified', 'Chromobacterium',
              'Lysinibacillus', 'Weissella', 'Corynebacterium', 'Cupriavidus',
              'Staphylococcus', 'Stenotrophomonas'],
    'Species': ['Unclassified', 'Unclassified', 'Unclassified', 'Unclassified',
                'boronitolerans', 'ghanensis', 'Unclassified', 'gilardii',
                'Unclassified', 'geniculata'],
    'Absolute Count': [3745, 10777, 4932, 1840, 1780, 1101, 703, 586, 568, 542]
}
df = pd.DataFrame(data)

def create_selective_label_pie(df, phylum_filter=None, genus_filter=None, species_filter=None):
    fig, ax = plt.subplots()
    size = 0.3

    total_phylum_counts = df.groupby('Phylum')['Absolute Count'].sum()
    phylum_colors = ['blue' if phylum in phylum_filter else 'none' for phylum in total_phylum_counts.index]
    phylum_labels = [phylum if phylum in phylum_filter else "" for phylum in total_phylum_counts.index]
    ax.pie(total_phylum_counts, labels=phylum_labels, colors=phylum_colors, radius=1,
           wedgeprops=dict(width=size, edgecolor='w'))

    total_genus_counts = df.groupby(['Phylum', 'Genus'])['Absolute Count'].sum()
    genus_colors = ['green' if (phylum in phylum_filter and genus in genus_filter) else 'none' 
                    for (phylum, genus) in total_genus_counts.index]
    genus_labels = [genus if (phylum in phylum_filter and genus in genus_filter) else ""
                    for (phylum, genus) in total_genus_counts.index]
    ax.pie(total_genus_counts, labels=genus_labels, colors=genus_colors, radius=1-size,
           wedgeprops=dict(width=size, edgecolor='w'), labeldistance=0.7)

    total_species_counts = df.groupby(['Phylum', 'Genus', 'Species'])['Absolute Count'].sum()
    species_colors = ['red' if (phylum in phylum_filter and genus in genus_filter and species in species_filter) else 'none'
                      for (phylum, genus, species) in total_species_counts.index]
    species_labels = [species if (phylum in phylum_filter and genus in genus_filter and species in species_filter) else ""
                      for (phylum, genus, species) in total_species_counts.index]
    ax.pie(total_species_counts, labels=species_labels, colors=species_colors, radius=1-2*size,
           wedgeprops=dict(width=size, edgecolor='w'), labeldistance=0.4)

    plt.title('Pie Chart with Selective Labels and Transparency')
    plt.show()

create_selective_label_pie(df, 
                           phylum_filter=['Firmicutes'],
                           genus_filter=['Weissella'],
                           species_filter=['boronitolerans'])

что дает (смехотворный пример, но показательный)

Этот

create_selective_label_pie(df, 
                           phylum_filter=['Proteobacteria', 'Firmicutes'],
                           genus_filter=['Lysinibacillus'],
                           species_filter=['boronitolerans'])

даст

Для цветовых вариаций вам придется что-то сделать самостоятельно.