�� : ��
��

import pandas as pd
import matplotlib.pyplot as plt
data = pd.read_csv('bacteria_list_200.csv')

family_frequency = data['Family'].value_counts(normalize=True)
family_table = family_frequency.to_frame(name='Frequency').reset_index()
family_table = family_table.rename(columns={'index': 'Family'})
threshold = 0.005
family_table['Family_cleaned'] = family_table.apply(lambda row: row['Family'] if row['Frequency'] > threshold else 'Other', axis=1)
cleaned_frequency = family_table.groupby('Family_cleaned')['Frequency'].sum().reset_index()
plt.figure(figsize=(10, 5))
plt.pie(cleaned_frequency['Frequency'], labels=cleaned_frequency['Family_cleaned'], autopct='%1.1f%%', startangle=140)
plt.axis('equal')
plt.title('���������� ����������� �������� ��������')
plt.show()
harmful_frequency = data['Harmful to Humans'].value_counts(normalize=True)
plt.figure(figsize=(5, 5))
plt.pie(harmful_frequency, labels=harmful_frequency.index, autopct='%1.1f%%', startangle=140)
plt.axis('equal')
plt.title('������������ � ��������� �������� ��� ��������')
plt.show()
�� -�� , � �� . � Pycharme ��
��
bacteria_list_200.csv
��
��������

��
