3.1 MiB
3.1 MiB
In [1]:
import pandas as pd
df = pd.read_csv("..//static//csv//heart_2020_cleaned.csv")
print(df.columns)
display(df.head(10))
Сегментация пациентов по рискам сердечно-сосудистых заболеваний Цель: Определить группы пациентов с различными уровнями риска развития сердечно-сосудистых заболеваний на основе их демографических данных, образа жизни и состояния здоровья.
Пример:
Кластер 1: Пациенты с высоким риском (курение, высокий ИМТ, низкая физическая активность).
Кластер 2: Пациенты со средним риском (умеренное курение, средний ИМТ, средняя физическая активность).
Кластер 3: Пациенты с низким риском (отсутствие вредных привычек, нормальный ИМТ, высокая физическая активность).
Бизнес-применение:
Разработка персонализированных программ профилактики и лечения для каждой группы пациентов.
Таргетированная реклама медицинских услуг и продуктов для улучшения образа жизни.
In [ ]:
import pandas as pd
df = pd.read_csv("..//static//csv//heart_2020_cleaned.csv").head(1000)
df = df.dropna()
from scipy import stats
z_scores = stats.zscore(df[['BMI', 'PhysicalHealth', 'MentalHealth', 'SleepTime']])
abs_z_scores = abs(z_scores)
filtered_entries = (abs_z_scores < 3).all(axis=1)
df = df[filtered_entries]
from sklearn.preprocessing import LabelEncoder
label_encoders = {}
for column in df.select_dtypes(include=['object']).columns:
le = LabelEncoder()
df[column] = le.fit_transform(df[column])
label_encoders[column] = le
from imblearn.over_sampling import RandomOverSampler
ros = RandomOverSampler(random_state=42)
X = df.drop('HeartDisease', axis=1)
y = df['HeartDisease']
X_resampled, y_resampled = ros.fit_resample(X, y)
df_balanced = X_resampled.copy()
df_balanced['HeartDisease'] = y_resampled
df = df_balanced
In [ ]:
scaler = StandardScaler()
df_encoded[numeric_features] = scaler.fit_transform(df_encoded[numeric_features])
In [ ]:
import matplotlib.pyplot as plt
import seaborn as sns
plt.figure(figsize=(16, 12))
plt.subplot(2, 2, 1)
sns.scatterplot(x='BMI', y='PhysicalHealth', hue='HeartDisease', data=df)
plt.title('BMI vs PhysicalHealth')
plt.subplot(2, 2, 2)
sns.scatterplot(x='MentalHealth', y='SleepTime', hue='HeartDisease', data=df)
plt.title('MentalHealth vs SleepTime')
plt.subplot(2, 2, 3)
sns.scatterplot(x='PhysicalHealth', y='SleepTime', hue='HeartDisease', data=df)
plt.title('PhysicalHealth vs SleepTime')
plt.subplot(2, 2, 4)
sns.scatterplot(x='BMI', y='MentalHealth', hue='HeartDisease', data=df)
plt.title('BMI vs MentalHealth')
plt.tight_layout()
plt.show()
In [ ]:
plt.figure(figsize=(16, 12))
plt.subplot(2, 2, 1)
sns.scatterplot(x='BMI', y='PhysicalHealth', data=df)
plt.title('BMI vs PhysicalHealth')
plt.subplot(2, 2, 2)
sns.scatterplot(x='MentalHealth', y='SleepTime', data=df)
plt.title('MentalHealth vs SleepTime')
plt.subplot(2, 2, 3)
sns.scatterplot(x='PhysicalHealth', y='SleepTime', data=df)
plt.title('PhysicalHealth vs SleepTime')
plt.subplot(2, 2, 4)
sns.scatterplot(x='BMI', y='MentalHealth', data=df)
plt.title('BMI vs MentalHealth')
plt.tight_layout()
plt.show()
In [ ]:
from sklearn.cluster import AgglomerativeClustering
from scipy.cluster.hierarchy import dendrogram, linkage
import matplotlib.pyplot as plt
X = df[['BMI', 'PhysicalHealth', 'MentalHealth', 'SleepTime']]
agg_clustering = AgglomerativeClustering(n_clusters=3)
clusters = agg_clustering.fit_predict(X)
df['Cluster'] = clusters
linked = linkage(X, 'ward')
plt.figure(figsize=(10, 7))
dendrogram(linked,
orientation='top',
distance_sort='descending',
show_leaf_counts=True)
plt.title('Dendrogram')
plt.show()
In [ ]:
import pandas as pd
from sklearn.preprocessing import LabelEncoder
from sklearn.cluster import MiniBatchKMeans
import matplotlib.pyplot as plt
def draw_data_2d(data, feature_x, feature_y, labels, subplot):
subplot.scatter(data.iloc[:, feature_x], data.iloc[:, feature_y], c=labels, cmap='viridis')
subplot.set_xlabel(data.columns[feature_x])
subplot.set_ylabel(data.columns[feature_y])
X = df[['BMI', 'PhysicalHealth', 'MentalHealth', 'SleepTime']]
kmeans = MiniBatchKMeans(n_clusters=3, random_state=0, batch_size=100)
result = kmeans.fit_predict(X)
plt.figure(figsize=(16, 24))
plt.subplot(4, 2, 1)
draw_data_2d(X, 0, 1, result, plt.subplot(4, 2, 1))
plt.title('Clusters (BMI vs PhysicalHealth)')
plt.subplot(4, 2, 2)
draw_data_2d(X, 0, 1, df['HeartDisease'], plt.subplot(4, 2, 2))
plt.title('True Labels (BMI vs PhysicalHealth)')
plt.subplot(4, 2, 3)
draw_data_2d(X, 2, 3, result, plt.subplot(4, 2, 3))
plt.title('Clusters (MentalHealth vs SleepTime)')
plt.subplot(4, 2, 4)
draw_data_2d(X, 2, 3, df['HeartDisease'], plt.subplot(4, 2, 4))
plt.title('True Labels (MentalHealth vs SleepTime)')
plt.subplot(4, 2, 5)
draw_data_2d(X, 0, 2, result, plt.subplot(4, 2, 5))
plt.title('Clusters (BMI vs MentalHealth)')
plt.subplot(4, 2, 6)
draw_data_2d(X, 0, 2, df['HeartDisease'], plt.subplot(4, 2, 6))
plt.title('True Labels (BMI vs MentalHealth)')
plt.subplot(4, 2, 7)
draw_data_2d(X, 1, 3, result, plt.subplot(4, 2, 7))
plt.title('Clusters (PhysicalHealth vs SleepTime)')
plt.subplot(4, 2, 8)
draw_data_2d(X, 1, 3, df['HeartDisease'], plt.subplot(4, 2, 8))
plt.title('True Labels (PhysicalHealth vs SleepTime)')
plt.tight_layout()
plt.show()
In [ ]:
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
def run_kmeans(data, n_clusters, random_state):
kmeans = KMeans(n_clusters=n_clusters, random_state=random_state)
labels = kmeans.fit_predict(data)
centers = kmeans.cluster_centers_
return labels, centers
def print_cluster_result(data, n_clusters, labels):
for i in range(n_clusters):
cluster_indices = [index for index, label in enumerate(labels) if label == i]
print(f"Cluster {i+1} ({len(cluster_indices)}):")
print(", ".join(map(str, cluster_indices)))
print("--------")
def draw_cluster_results(data, feature_x, feature_y, labels, centers, subplot):
subplot.scatter(data.iloc[:, feature_x], data.iloc[:, feature_y], c=labels, cmap='viridis')
subplot.scatter(centers[:, feature_x], centers[:, feature_y], marker='x', s=200, linewidths=3, color='r')
subplot.set_xlabel(data.columns[feature_x])
subplot.set_ylabel(data.columns[feature_y])
random_state = 9
n_clusters = 3
labels, centers = run_kmeans(X, n_clusters, random_state)
print_cluster_result(X, n_clusters, labels)
display(centers)
display(df['HeartDisease'])
plt.figure(figsize=(16, 12))
draw_cluster_results(X, 0, 1, labels, centers, plt.subplot(2, 2, 1))
draw_cluster_results(X, 2, 3, labels, centers, plt.subplot(2, 2, 2))
draw_cluster_results(X, 0, 2, labels, centers, plt.subplot(2, 2, 3))
draw_cluster_results(X, 1, 3, labels, centers, plt.subplot(2, 2, 4))
plt.tight_layout()
plt.show()
In [ ]:
from sklearn.decomposition import PCA
reduced_data = PCA(n_components=2).fit_transform(X)
print(reduced_data)
In [ ]:
import matplotlib.pyplot as plt
import pandas as pd
plt.figure(figsize=(16, 6))
draw_data_2d(
pd.DataFrame({"Column1": reduced_data[:, 0], "Column2": reduced_data[:, 1]}),
0,
1,
labels=[0] * len(reduced_data),
subplot=plt.subplot(1, 2, 1),
)
plt.title('PCA Data (Without True Labels)')
draw_data_2d(
pd.DataFrame({"Column1": reduced_data[:, 0], "Column2": reduced_data[:, 1]}),
0,
1,
df['HeartDisease'],
plt.subplot(1, 2, 2),
)
plt.title('PCA Data (With True Labels)')
plt.tight_layout()
plt.show()
In [ ]:
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
def fit_kmeans(data, n_clusters, random_state):
kmeans = KMeans(n_clusters=n_clusters, random_state=random_state)
kmeans.fit(data)
return kmeans
def draw_clusters(data, kmeans):
labels = kmeans.labels_
centers = kmeans.cluster_centers_
plt.figure(figsize=(10, 6))
plt.scatter(data[:, 0], data[:, 1], c=labels, cmap='viridis', alpha=0.6)
plt.scatter(centers[:, 0], centers[:, 1], marker='x', s=200, linewidths=3, color='r')
plt.xlabel('Principal Component 1')
plt.ylabel('Principal Component 2')
plt.title('K-means Clustering (2 Clusters) with PCA')
plt.show()
kmeans = fit_kmeans(reduced_data, 2, random_state)
draw_clusters(reduced_data, kmeans)
In [ ]:
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
def fit_kmeans(data, n_clusters, random_state):
kmeans = KMeans(n_clusters=n_clusters, random_state=random_state)
kmeans.fit(data)
return kmeans
def draw_clusters(data, kmeans):
labels = kmeans.labels_
centers = kmeans.cluster_centers_
plt.figure(figsize=(10, 6))
plt.scatter(data[:, 0], data[:, 1], c=labels, cmap='viridis', alpha=0.6)
plt.scatter(centers[:, 0], centers[:, 1], marker='x', s=200, linewidths=3, color='white')
plt.xlabel('Principal Component 1')
plt.ylabel('Principal Component 2')
plt.title('K-means Clustering (PCA-reduced data)')
plt.show()
kmeans = fit_kmeans(reduced_data, 3, random_state)
draw_clusters(reduced_data, kmeans)
In [ ]:
import matplotlib.pyplot as plt
import pandas as pd
labels = [2 if val == 1 else 1 if val == 2 else val for val in kmeans.labels_]
plt.figure(figsize=(16, 6))
draw_data_2d(
pd.DataFrame({"Column1": reduced_data[:, 0], "Column2": reduced_data[:, 1]}),
0,
1,
labels,
plt.subplot(1, 2, 1),
)
plt.title('Clusters (PCA-reduced data)')
draw_data_2d(
pd.DataFrame({"Column1": reduced_data[:, 0], "Column2": reduced_data[:, 1]}),
0,
1,
df['HeartDisease'],
plt.subplot(1, 2, 2),
)
plt.title('True Labels (PCA-reduced data)')
plt.tight_layout()
plt.show()
In [ ]:
from sklearn.decomposition import PCA
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
def get_clusters_inertia(data, random_state, max_clusters=10):
inertias = []
clusters_range = range(1, max_clusters + 1)
for n_clusters in clusters_range:
kmeans = KMeans(n_clusters=n_clusters, random_state=random_state)
kmeans.fit(data)
inertias.append(kmeans.inertia_)
return inertias, clusters_range
def draw_elbow_diagram(inertias, clusters_range):
plt.figure(figsize=(10, 6))
plt.plot(clusters_range, inertias, marker='o')
plt.xlabel('Number of Clusters')
plt.ylabel('Inertia')
plt.title('Elbow Method for Optimal Number of Clusters')
plt.show()
random_state = 42
max_clusters = 10
inertias, clusters_range = get_clusters_inertia(reduced_data, random_state, max_clusters)
print("Clusters Range:", list(clusters_range))
print("Inertias:", inertias)
draw_elbow_diagram(inertias, clusters_range)
In [30]:
from sklearn.metrics import silhouette_score
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt
import numpy as np
def get_clusters_silhouette_scores(data, random_state, max_clusters=10):
silhouette_scores = []
clusters_range = range(2, max_clusters + 1)
for n_clusters in clusters_range:
kmeans = KMeans(n_clusters=n_clusters, random_state=random_state)
labels = kmeans.fit_predict(data)
score = silhouette_score(data, labels)
silhouette_scores.append(score)
return silhouette_scores, clusters_range
def draw_silhouettes_diagram(silhouette_scores, clusters_range):
plt.figure(figsize=(10, 6))
plt.plot(clusters_range, silhouette_scores, marker='o')
plt.xlabel('Number of Clusters')
plt.ylabel('Silhouette Score')
plt.title('Silhouette Analysis for Optimal Number of Clusters')
plt.show()
random_state = 42
max_clusters = 10
scaler = StandardScaler()
reduced_data = scaler.fit_transform(reduced_data)
silhouette_scores, clusters_range = get_clusters_silhouette_scores(reduced_data, random_state, max_clusters)
print("Clusters Range:", list(clusters_range))
print("Silhouette Scores:", silhouette_scores)
draw_silhouettes_diagram(silhouette_scores, clusters_range)
In [33]:
import math
import matplotlib.pyplot as plt
from sklearn.metrics import silhouette_samples, silhouette_score
from sklearn.cluster import KMeans
import numpy as np
def get_clusters_silhouettes(data, random_state):
max_clusters = min(12, int(math.sqrt(len(data))))
silhouettes = []
for n_clusters in range(2, max_clusters + 1):
kmeans = KMeans(n_clusters=n_clusters, random_state=random_state)
cluster_labels = kmeans.fit_predict(data)
silhouette_avg = silhouette_score(data, cluster_labels)
silhouettes.append((n_clusters, silhouette_avg, cluster_labels, kmeans.cluster_centers_))
return silhouettes
def draw_silhouettes(data, silhouettes):
for n_clusters, silhouette_avg, cluster_labels, centers in silhouettes:
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(15, 6))
ax1.set_xlim([-0.1, 1])
ax1.set_ylim([0, len(data) + (n_clusters + 1) * 10])
silhouette_values = silhouette_samples(data, cluster_labels)
y_lower = 10
for i in range(n_clusters):
cluster_silhouette_values = silhouette_values[cluster_labels == i]
cluster_silhouette_values.sort()
cluster_size = cluster_silhouette_values.shape[0]
y_upper = y_lower + cluster_size
ax1.fill_betweenx(np.arange(y_lower, y_upper), 0, cluster_silhouette_values, alpha=0.7)
ax1.text(-0.05, y_lower + 0.5 * cluster_size, str(i))
y_lower = y_upper + 10
ax1.set_title("Silhouette plot for {} clusters".format(n_clusters))
ax1.set_xlabel("Silhouette coefficient values")
ax1.set_ylabel("Cluster label")
ax1.axvline(x=silhouette_avg, color="red", linestyle="--")
ax1.set_yticks([])
ax2.scatter(data[:, 0], data[:, 1], c=cluster_labels, cmap='viridis', s=30)
ax2.scatter(centers[:, 0], centers[:, 1], marker='x', c='red', s=200, alpha=1)
ax2.set_title("Clustered data for {} clusters".format(n_clusters))
ax2.set_xlabel("Feature space for the 1st feature")
ax2.set_ylabel("Feature space for the 2nd feature")
plt.suptitle(("Silhouette analysis for KMeans clustering with {} clusters".format(n_clusters)),
fontsize=14, fontweight='bold')
plt.show()
random_state = 42
silhouettes = get_clusters_silhouettes(reduced_data, random_state)
draw_silhouettes(reduced_data, silhouettes)
