التعلم غير الموجّه: Clustering

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler

rng = np.random.default_rng(0)
customers = pd.DataFrame({
    "visits": np.concatenate([rng.normal(3, 1, 80), rng.normal(12, 2, 60),
                              rng.normal(6, 1.5, 60)]).clip(1),
    "spend": np.concatenate([rng.normal(400, 120, 80), rng.normal(2500, 400, 60),
                             rng.normal(5200, 700, 60)]).clip(50),
})

X = StandardScaler().fit_transform(customers)      # distances need same scale!
kmeans = KMeans(n_clusters=3, n_init=10, random_state=0).fit(X)
customers["segment"] = kmeans.labels_

print(customers.groupby("segment").mean().round(0))

plt.figure(figsize=(7, 3.8))
plt.scatter(customers["visits"], customers["spend"], c=customers["segment"],
            cmap="viridis", s=25)
plt.xlabel("visits per month")
plt.ylabel("monthly spend (EGP)")
plt.title("Customer segments found by K-Means")
plt.show()

import numpy as np
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler

rng = np.random.default_rng(0)
visits = np.concatenate([rng.normal(3, 1, 80), rng.normal(12, 2, 60), rng.normal(6, 1.5, 60)])
spend = np.concatenate([rng.normal(400, 120, 80), rng.normal(2500, 400, 60), rng.normal(5200, 700, 60)])
X = StandardScaler().fit_transform(np.column_stack([visits, spend]))

ks = range(1, 9)
inertia = [KMeans(n_clusters=k, n_init=10, random_state=0).fit(X).inertia_ for k in ks]

plt.figure(figsize=(7, 3.4))
plt.plot(ks, inertia, marker="o", color="#3776AB")
plt.xlabel("number of clusters (K)")
plt.ylabel("inertia (lower = tighter)")
plt.title("Elbow method")
plt.grid(alpha=0.3)
plt.show()