import matplotlib.pyplot as plt import numpy as np import pandas as pd import seaborn as sns from scipy.cluster.vq import kmeans, kmeans2, vq from sklearn.cluster import KMeans from sklearn.metrics import silhouette_score from yellowbrick.cluster import SilhouetteVisualizer plt.style.use("fivethirtyeight")

data = pd.read_csv('data/data.csv') data.head()

sns.scatterplot(x='x_scaled', y='y_scaled', data=data) plt.title("Jelas Terlihat Terdapat 2 Cluster", loc="left") plt.grid(False) plt.show()

centroids, labels = kmeans2(data[["x_scaled", "y_scaled"]], k=2, minit="++") print("centroids:", centroids, sep="\n") print("labels:", labels) data["labels_kmeans2"] = labels sns.scatterplot(x='x_scaled', y='y_scaled', data=data, hue="labels_kmeans2") plt.scatter(centroids[:, 0], centroids[:, 1], marker="*", c="black", label="centroid") plt.title("Implementasi kmeans2", loc="left") plt.legend() plt.grid(False) plt.show()

data[["x", "y", "labels_kmeans2"]]

# Inisialisasi pusat klaster cluster_centers, distortion = kmeans(data[['x_scaled', 'y_scaled']], k_or_guess=2) print("centroids:", cluster_centers, sep="\n") print("distortion:", distortion) # Penetapan label klaster data['labels_kmeans'], distortion_list = vq(data[['x_scaled', 'y_scaled']], cluster_centers) print(f"{distortion_list = }\n{distortion_list.mean() = }") print("labels:", data["labels_kmeans"].tolist()) # Plot clusters sns.scatterplot(x='x_scaled', y='y_scaled', hue='labels_kmeans', data=data) plt.scatter(cluster_centers[:, 0], cluster_centers[:, 1], marker="*", c="black", label="centroid") plt.title("Implementasi kmeans", loc="left") plt.legend() plt.grid(False) plt.show()

data[["x", "y", "labels_kmeans2", "labels_kmeans"]]

k_means_2 = KMeans(n_clusters=2) k_means_2.fit(data[["x_scaled", "y_scaled"]]) centroids = k_means_2.cluster_centers_ inertia = k_means_2.inertia_ print("centroids:", centroids, sep="\n") print("inertia:", inertia) labels = k_means_2.predict(data[["x_scaled", "y_scaled"]]) data["labels_sklearn"] = labels print("labels:", labels) sns.scatterplot(x='x_scaled', y='y_scaled', hue='labels_sklearn', data=data) plt.scatter(centroids[:, 0], centroids[:, 1], marker="*", c="black", label="centroid") plt.title("Implementasi kmeans", loc="left") plt.legend() plt.grid(False) plt.show()

cluster_0 = data.loc[data["labels_sklearn"] == 0, ["x", "y"]] cluster_1 = data.loc[data["labels_sklearn"] == 1, ["x", "y"]]

cluster_0.describe().T

cluster_1.describe().T

k_means_3 = KMeans(n_clusters=3) k_means_3.fit(data[["x_scaled", "y_scaled"]]) centroids = k_means_3.cluster_centers_ inertia = k_means_3.inertia_ print("centroids:", centroids, sep="\n") print("inertia:", inertia) labels = k_means_3.predict(data[["x_scaled", "y_scaled"]]) data["labels_sklearn"] = labels print("labels:", labels) sns.scatterplot(x='x_scaled', y='y_scaled', hue='labels_sklearn', data=data) plt.scatter(centroids[:, 0], centroids[:, 1], marker="*", c="black", label="centroid") plt.title("Implementasi kmeans", loc="left") plt.legend() plt.grid(False) plt.show()

cluster_0 = data.loc[data["labels_sklearn"] == 0, ["x", "y"]] cluster_1 = data.loc[data["labels_sklearn"] == 1, ["x", "y"]] cluster_2 = data.loc[data["labels_sklearn"] == 2, ["x", "y"]]

display(cluster_0.describe().T) display(cluster_1.describe().T) display(cluster_2.describe().T)

distortions = [] num_clusters = range(1, 7) # Buat list untuk distorsi dari kmeans for i in num_clusters: cluster_centers, distortion = kmeans(data[['x_scaled', 'y_scaled']], k_or_guess=i) distortions.append(distortion) # Buat dataframe dengan dua lists - num_clusters, distortions elbow_plot = pd.DataFrame({'num_clusters': num_clusters, 'distortions': distortions}) # Buat line plotnya sns.lineplot(x='num_clusters', y='distortions', data = elbow_plot) plt.xticks(num_clusters) plt.title("Elbow method using distortion") plt.grid(False) plt.show()

ssd = [] clusters = range(1, 11) for k in clusters: k_means = KMeans(n_clusters=k, random_state=11) k_means.fit(data[["x_scaled", "y_scaled"]]) ssd.append(k_means.inertia_) plt.plot(clusters, ssd) plt.xlabel("Clusters") plt.ylabel("Inertia") plt.title("Elbow method using inertia", loc="left") plt.grid(False) plt.show()

k_means_2 = KMeans(n_clusters=2) k_means_2.fit(data[["x_scaled", "y_scaled"]]) labels_2 = k_means_2.predict(data[["x_scaled", "y_scaled"]]) average_silhouette_score_2 = silhouette_score( data[["x_scaled", "y_scaled"]], labels_2 ) print(f"(k=2) {average_silhouette_score_2 = }") k_means_3 = KMeans(n_clusters=3) k_means_3.fit(data[["x_scaled", "y_scaled"]]) labels_3 = k_means_3.predict(data[["x_scaled", "y_scaled"]]) average_silhouette_score_3 = silhouette_score( data[["x_scaled", "y_scaled"]], labels_3 ) print(f"(k=3) {average_silhouette_score_3 = }")

silhouette_viz = SilhouetteVisualizer(k_means_2) silhouette_viz.fit(data[["x_scaled", "y_scaled"]]) plt.grid(False) silhouette_viz.show() plt.show()

silhouette_viz = SilhouetteVisualizer(k_means_3) silhouette_viz.fit(data[["x_scaled", "y_scaled"]]) plt.grid(False) silhouette_viz.show() plt.show()