AutoML и NAS / Comments / Habr

import psycopg2

import pandas as pd

from sklearn.cluster import KMeans

from sklearn.model_selection import train_test_split

from sklearn.metrics import silhouette_score, classification_report, confusion_matrix

from sklearn.preprocessing import StandardScaler

import seaborn as sns

import matplotlib.pyplot as plt

# Подключение к базе данных

def connect_to_db():

conn = psycopg2.connect(host="localhost", database="primer", user="postgres", password="12345")

return conn, conn.cursor()

# Загрузка данных

def fetch_data(cursor, table_name="station_data_4"):

cursor.execute(f"SELECT * FROM {table_name}")

return pd.DataFrame(cursor.fetchall(), columns=[desc[0] for desc in cursor.description])

# Основная функция

def main():

conn, cursor = connect_to_db()

if conn and cursor:

df = fetch_data(cursor)

if df is not None:

X = df[['travel_time_interval_average', 'station_load_total', 'capacity_percent_avg', 'peak_travel_duration']]

X_scaled = StandardScaler().fit_transform(X)

kmeans = KMeans(n_clusters=3, random_state=42)

df['Cluster'] = kmeans.fit_predict(X_scaled)

X_train, X_test, y_train, y_test = train_test_split(X_scaled, df['Cluster'], test_size=0.3, random_state=42)

labels_train = kmeans.fit_predict(X_train)

labels_test = kmeans.predict(X_test)

print("Silhouette Score:", silhouette_score(X_train, labels_train))

print("Classification Report:\n", classification_report(y_test, labels_test))

print("Confusion Matrix:\n", confusion_matrix(y_test, labels_test))

# Создаем DataFrame для визуализации

df_visualize = pd.DataFrame(X_train, columns=X.columns)

df_visualize['Cluster'] = labels_train # Добавляем столбец с кластерами

# Визуализация

sns.pairplot(df_visualize, hue='Cluster', palette='Set2')

plt.show()

cursor.close()

conn.close()

if name == "main":

main()

AutoML и NAS