inertia = []<\/p>\nfor i in range(1, 11):<\/p>\n
kmeans = KMeans(n_clusters=i, random_state=0)<\/p>\n
kmeans.fit(data_scaled)<\/p>\n
inertia.append(kmeans.inertia_)<\/p>\n
plt.plot(range(1, 11), inertia)<\/p>\n
plt.title('Elbow Method')<\/p>\n
plt.xlabel('Number of Clusters')<\/p>\n
plt.ylabel('Inertia')<\/p>\n
plt.show()<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u901a\u8fc7\u8098\u90e8\u6cd5\u5219\uff0c\u6211\u4eec\u53ef\u4ee5\u89c2\u5bdf\u5230\u4e00\u4e2a\u62d0\u70b9\uff0c\u62d0\u70b9\u5bf9\u5e94\u7684\u7c07\u6570\u901a\u5e38\u662f\u8f83\u597d\u7684\u9009\u62e9\u3002<\/p>\n<\/p>\n
2. \u521d\u59cb\u5316\u65b9\u5f0f<\/h3>\n<\/p>\n
KMeans\u7b97\u6cd5\u7684\u521d\u59cb\u5316\u53ef\u4ee5\u9009\u62e9\u4e0d\u540c\u7684\u7b56\u7565\uff0c\u5e38\u89c1\u7684\u6709\u201ck-means++\u201d\u548c\u201c\u968f\u673a\u521d\u59cb\u5316\u201d\u3002\u201ck-means++\u201d\u901a\u5e38\u80fd\u5f97\u5230\u66f4\u597d\u7684\u805a\u7c7b\u6548\u679c\uff1a<\/p>\n<\/p>\n
kmeans = KMeans(n_clusters=3, init='k-means++', random_state=0)<\/p>\n<\/code><\/pre>\n<\/p>\n
\u56db\u3001\u62df\u5408KMeans\u6a21\u578b<\/h2>\n<\/p>\n
\u5728\u521d\u59cb\u5316\u6a21\u578b\u540e\uff0c\u9700\u8981\u62df\u5408\u6a21\u578b\u4ee5\u8fdb\u884c\u805a\u7c7b\u5206\u6790\u3002\u5728scikit-learn\u4e2d\uff0c\u8fd9\u4e00\u6b65\u901a\u8fc7fit<\/code>\u65b9\u6cd5\u6765\u5b9e\u73b0\u3002<\/p>\n<\/p>\nkmeans.fit(data_scaled)<\/p>\n<\/code><\/pre>\n<\/p>\n
\u62df\u5408\u6a21\u578b\u540e\uff0c\u6211\u4eec\u53ef\u4ee5\u83b7\u5f97\u6bcf\u4e2a\u6570\u636e\u70b9\u7684\u805a\u7c7b\u6807\u7b7e\uff1a<\/p>\n<\/p>\n
labels = kmeans.labels_<\/p>\n<\/code><\/pre>\n<\/p>\n
\u8fd8\u53ef\u4ee5\u83b7\u53d6\u6a21\u578b\u7684\u805a\u5408\u5ea6\uff08inertia\uff09\uff0c\u7528\u4e8e\u8bc4\u4f30\u6a21\u578b\u7684\u597d\u574f\uff1a<\/p>\n<\/p>\n
inertia = kmeans.inertia_<\/p>\n<\/code><\/pre>\n<\/p>\n
\u4e94\u3001\u83b7\u53d6\u805a\u7c7b\u7ed3\u679c<\/h2>\n<\/p>\n
\u901a\u8fc7KMeans\u6a21\u578b\u7684\u62df\u5408\uff0c\u6211\u4eec\u53ef\u4ee5\u83b7\u53d6\u6bcf\u4e2a\u6570\u636e\u70b9\u7684\u805a\u7c7b\u6807\u7b7e\u548c\u805a\u7c7b\u4e2d\u5fc3\u3002\u8fd9\u4e9b\u4fe1\u606f\u5bf9\u4e8e\u7406\u89e3\u6570\u636e\u7684\u5206\u5e03\u548c\u805a\u7c7b\u6548\u679c\u975e\u5e38\u91cd\u8981\u3002<\/p>\n<\/p>\n
1. \u805a\u7c7b\u6807\u7b7e<\/h3>\n<\/p>\n
\u805a\u7c7b\u6807\u7b7e\u901a\u8fc7\u6a21\u578b\u7684labels_<\/code>\u5c5e\u6027\u83b7\u53d6\uff0c\u5b83\u8868\u793a\u6bcf\u4e2a\u6570\u636e\u70b9\u6240\u5c5e\u7684\u7c07\uff1a<\/p>\n<\/p>\nlabels = kmeans.labels_<\/p>\n<\/code><\/pre>\n<\/p>\n
\u53ef\u4ee5\u5c06\u8fd9\u4e9b\u6807\u7b7e\u6dfb\u52a0\u5230\u539f\u59cb\u6570\u636e\u4e2d\uff0c\u4ee5\u4fbf\u66f4\u597d\u5730\u8fdb\u884c\u540e\u7eed\u5206\u6790\uff1a<\/p>\n<\/p>\n
data['Cluster'] = labels<\/p>\n<\/code><\/pre>\n<\/p>\n
2. \u805a\u7c7b\u4e2d\u5fc3<\/h3>\n<\/p>\n
\u805a\u7c7b\u4e2d\u5fc3\u901a\u8fc7\u6a21\u578b\u7684cluster_centers_<\/code>\u5c5e\u6027\u83b7\u53d6\uff0c\u8868\u793a\u6bcf\u4e2a\u7c07\u7684\u4e2d\u5fc3\u70b9\uff1a<\/p>\n<\/p>\ncenters = kmeans.cluster_centers_<\/p>\n<\/code><\/pre>\n<\/p>\n
\u8fd9\u4e9b\u4e2d\u5fc3\u70b9\u53ef\u4ee5\u7528\u4e8e\u8fdb\u4e00\u6b65\u5206\u6790\u805a\u7c7b\u7279\u5f81\u3002<\/p>\n<\/p>\n
\u516d\u3001\u53ef\u89c6\u5316\u805a\u7c7b\u7ed3\u679c<\/h2>\n<\/p>\n
\u53ef\u89c6\u5316\u662f\u7406\u89e3\u805a\u7c7b\u6548\u679c\u7684\u91cd\u8981\u624b\u6bb5\u3002\u901a\u8fc7Matplotlib\u5e93\uff0c\u6211\u4eec\u53ef\u4ee5\u5f88\u65b9\u4fbf\u5730\u5c06\u805a\u7c7b\u7ed3\u679c\u53ef\u89c6\u5316\u3002<\/p>\n<\/p>\n
1. \u7ed8\u5236\u805a\u7c7b\u7ed3\u679c<\/h3>\n<\/p>\n
\u4ee5\u4e0b\u662f\u4e00\u4e2a\u7b80\u5355\u7684\u53ef\u89c6\u5316\u805a\u7c7b\u7ed3\u679c\u7684\u4f8b\u5b50\uff1a<\/p>\n<\/p>\n
plt.scatter(data_scaled[:, 0], data_scaled[:, 1], c=labels, cmap='viridis')<\/p>\nplt.scatter(centers[:, 0], centers[:, 1], c='red', marker='x', s=200, alpha=0.75)<\/p>\n
plt.title('KMeans Clustering')<\/p>\n
plt.xlabel('Feature 1')<\/p>\n
plt.ylabel('Feature 2')<\/p>\n
plt.show()<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u8fd9\u79cd\u53ef\u89c6\u5316\u65b9\u6cd5\u53ef\u4ee5\u5e2e\u52a9\u6211\u4eec\u76f4\u89c2\u5730\u770b\u5230\u6570\u636e\u88ab\u5206\u6210\u4e86\u51e0\u4e2a\u7c07\uff0c\u4ee5\u53ca\u6bcf\u4e2a\u7c07\u7684\u4e2d\u5fc3\u4f4d\u7f6e\u3002<\/p>\n<\/p>\n
2. \u4e0d\u540c\u7ef4\u5ea6\u7684\u53ef\u89c6\u5316<\/h3>\n<\/p>\n
\u5982\u679c\u6570\u636e\u7ef4\u5ea6\u8f83\u9ad8\uff0c\u53ef\u4ee5\u4f7f\u7528\u4e3b\u6210\u5206\u5206\u6790\uff08PCA\uff09\u6216t-SNE\u7b49\u964d\u7ef4\u6280\u672f\uff0c\u5c06\u6570\u636e\u964d\u5230\u4e8c\u7ef4\u518d\u8fdb\u884c\u53ef\u89c6\u5316\uff1a<\/p>\n<\/p>\n
from sklearn.decomposition import PCA<\/p>\npca = PCA(n_components=2)<\/p>\n
data_pca = pca.fit_transform(data_scaled)<\/p>\n
plt.scatter(data_pca[:, 0], data_pca[:, 1], c=labels, cmap='viridis')<\/p>\n
plt.title('PCA of KMeans Clustering')<\/p>\n
plt.xlabel('Principal Component 1')<\/p>\n
plt.ylabel('Principal Component 2')<\/p>\n
plt.show()<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u901a\u8fc7\u4e0d\u540c\u7ef4\u5ea6\u7684\u53ef\u89c6\u5316\uff0c\u6211\u4eec\u53ef\u4ee5\u83b7\u5f97\u6570\u636e\u5728\u4e0d\u540c\u7279\u5f81\u7a7a\u95f4\u4e2d\u7684\u805a\u7c7b\u6548\u679c\u3002<\/p>\n<\/p>\n
\u4e03\u3001\u4f18\u5316\u6a21\u578b\u53c2\u6570<\/h2>\n<\/p>\n
\u5728\u521d\u6b65\u5b8c\u6210KMeans\u805a\u7c7b\u5206\u6790\u540e\uff0c\u901a\u5e38\u9700\u8981\u5bf9\u6a21\u578b\u53c2\u6570\u8fdb\u884c\u4f18\u5316\uff0c\u4ee5\u63d0\u9ad8\u805a\u7c7b\u6548\u679c\u3002<\/p>\n<\/p>\n
1. \u9009\u62e9\u6700\u4f73\u7684\u7c07\u6570<\/h3>\n<\/p>\n
\u524d\u9762\u63d0\u5230\u7684\u201c\u8098\u90e8\u6cd5\u5219\u201d\u53ef\u4ee5\u7528\u6765\u9009\u62e9\u6700\u4f73\u7684\u7c07\u6570\u3002\u6b64\u5916\uff0c\u8fd8\u53ef\u4ee5\u4f7f\u7528\u201c\u8f6e\u5ed3\u7cfb\u6570\u201d\u7b49\u5176\u4ed6\u6307\u6807\u8fdb\u884c\u8bc4\u4f30\uff1a<\/p>\n<\/p>\n
from sklearn.metrics import silhouette_score<\/p>\nsilhouette_avg = silhouette_score(data_scaled, labels)<\/p>\n
print(f'Silhouette Score: {silhouette_avg}')<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u8f6e\u5ed3\u7cfb\u6570\u8d8a\u63a5\u8fd11\uff0c\u8868\u660e\u805a\u7c7b\u6548\u679c\u8d8a\u597d\u3002<\/p>\n<\/p>\n
2. \u8c03\u6574\u521d\u59cb\u5316\u53c2\u6570<\/h3>\n<\/p>\n
\u9664\u4e86\u7c07\u6570\uff0c\u521d\u59cb\u5316\u53c2\u6570\uff08\u5982init<\/code>\u548cn_init<\/code>\uff09\u4e5f\u4f1a\u5f71\u54cd\u805a\u7c7b\u6548\u679c\u3002\u53ef\u4ee5\u5c1d\u8bd5\u4e0d\u540c\u7684\u521d\u59cb\u5316\u7b56\u7565\u548c\u6b21\u6570\uff1a<\/p>\n<\/p>\nkmeans = KMeans(n_clusters=3, init='random', n_init=10, random_state=0)<\/p>\nkmeans.fit(data_scaled)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u901a\u8fc7\u8c03\u6574\u8fd9\u4e9b\u53c2\u6570\uff0c\u53ef\u4ee5\u8fdb\u4e00\u6b65\u63d0\u9ad8\u6a21\u578b\u7684\u7a33\u5b9a\u6027\u548c\u51c6\u786e\u6027\u3002<\/p>\n<\/p>\n
3. \u5904\u7406\u5f02\u5e38\u6570\u636e<\/h3>\n<\/p>\n
\u5f02\u5e38\u6570\u636e\u53ef\u80fd\u4f1a\u5f71\u54cd\u805a\u7c7b\u6548\u679c\u3002\u5728\u8fdb\u884cKMeans\u805a\u7c7b\u5206\u6790\u524d\uff0c\u53ef\u4ee5\u4f7f\u7528\u5f02\u5e38\u68c0\u6d4b\u65b9\u6cd5\u53bb\u9664\u5f02\u5e38\u6570\u636e\uff1a<\/p>\n<\/p>\n
from sklearn.ensemble import IsolationForest<\/p>\niso = IsolationForest(contamination=0.1)<\/p>\n
yhat = iso.fit_predict(data_scaled)<\/p>\n
mask = yhat != -1<\/p>\n
data_clean = data_scaled[mask]<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u901a\u8fc7\u53bb\u9664\u5f02\u5e38\u6570\u636e\uff0c\u53ef\u4ee5\u63d0\u9ad8\u805a\u7c7b\u6a21\u578b\u7684\u6027\u80fd\u3002<\/p>\n<\/p>\n
\n\u901a\u8fc7\u4e0a\u8ff0\u6b65\u9aa4\uff0c\u6211\u4eec\u53ef\u4ee5\u6709\u6548\u5730\u5728Python\u4e2d\u4f7f\u7528KMeans\u7b97\u6cd5\u8fdb\u884c\u805a\u7c7b\u5206\u6790\u3002\u6839\u636e\u5177\u4f53\u7684\u6570\u636e\u548c\u4e1a\u52a1\u9700\u6c42\uff0c\u8fd8\u53ef\u4ee5\u8fdb\u884c\u8fdb\u4e00\u6b65\u7684\u6a21\u578b\u4f18\u5316\u548c\u7ed3\u679c\u89e3\u91ca\u3002\u5e0c\u671b\u8fd9\u4e9b\u6b65\u9aa4\u548c\u65b9\u6cd5\u80fd\u591f\u5e2e\u52a9\u60a8\u66f4\u597d\u5730\u7406\u89e3\u548c\u5e94\u7528KMeans\u805a\u7c7b\u7b97\u6cd5\u3002<\/p>\n<\/p>\n
\u76f8\u5173\u95ee\u7b54FAQs\uff1a<\/strong><\/h2>\n \u5982\u4f55\u5728Python\u4e2d\u5b9e\u73b0KMeans\u805a\u7c7b\u7b97\u6cd5\uff1f<\/strong>
KMeans\u662f\u4e00\u79cd\u5e38\u7528\u7684\u805a\u7c7b\u7b97\u6cd5\uff0c\u60a8\u53ef\u4ee5\u901a\u8fc7\u4f7f\u7528scikit-learn<\/code>\u5e93\u8f7b\u677e\u5b9e\u73b0\u3002\u9996\u5148\uff0c\u786e\u4fdd\u5b89\u88c5\u4e86\u8fd9\u4e2a\u5e93\u3002\u60a8\u53ef\u4ee5\u901a\u8fc7\u547d\u4ee4pip install scikit-learn<\/code>\u8fdb\u884c\u5b89\u88c5\u3002\u63a5\u4e0b\u6765\uff0c\u5bfc\u5165\u6240\u9700\u7684\u6a21\u5757\uff0c\u51c6\u5907\u6570\u636e\uff0c\u9009\u62e9\u805a\u7c7b\u7684\u6570\u91cf\uff08k\u503c\uff09\uff0c\u7136\u540e\u4f7f\u7528KMeans\u7c7b\u8fdb\u884c\u62df\u5408\uff0c\u6700\u540e\u53ef\u4ee5\u901a\u8fc7predict<\/code>\u65b9\u6cd5\u8fdb\u884c\u9884\u6d4b\u3002<\/p>\nKMeans\u7b97\u6cd5\u7684\u9002\u7528\u573a\u666f\u6709\u54ea\u4e9b\uff1f<\/strong>
KMeans\u7b97\u6cd5\u9002\u7528\u4e8e\u8bb8\u591a\u573a\u666f\uff0c\u4f8b\u5982\u5e02\u573a\u7ec6\u5206\u3001\u793e\u4ea4\u7f51\u7edc\u5206\u6790\u3001\u56fe\u50cf\u538b\u7f29\u548c\u5f02\u5e38\u68c0\u6d4b\u7b49\u3002\u5b83\u7279\u522b\u9002\u5408\u5904\u7406\u5927\u89c4\u6a21\u6570\u636e\u96c6\uff0c\u5e76\u4e14\u5bf9\u4e8e\u7403\u5f62\u5206\u5e03\u7684\u6570\u636e\u6548\u679c\u6700\u4f73\u3002\u60a8\u53ef\u4ee5\u5229\u7528KMeans\u6765\u8bc6\u522b\u6570\u636e\u4e2d\u7684\u81ea\u7136\u5206\u7ec4\uff0c\u5e2e\u52a9\u60a8\u66f4\u597d\u5730\u7406\u89e3\u6570\u636e\u7ed3\u6784\u3002<\/p>\n\u5982\u4f55\u9009\u62e9KMeans\u4e2d\u7684k\u503c\uff1f<\/strong>
\u9009\u62e9\u5408\u9002\u7684k\u503c\u662fKMeans\u805a\u7c7b\u4e2d\u7684\u4e00\u4e2a\u5173\u952e\u6b65\u9aa4\u3002\u5e38\u89c1\u7684\u65b9\u6cd5\u5305\u62ec\u8098\u90e8\u6cd5\u5219\uff08Elbow Method\uff09\u548c\u8f6e\u5ed3\u7cfb\u6570\uff08Silhouette Score\uff09\u3002\u8098\u90e8\u6cd5\u5219\u901a\u8fc7\u7ed8\u5236\u4e0d\u540ck\u503c\u4e0b\u7684\u805a\u7c7b\u6210\u672c\uff08\u5982SSE\uff09\u56fe\u5f62\uff0c\u5bfb\u627e\u201c\u8098\u90e8\u201d\u70b9\u6765\u786e\u5b9a\u6700\u4f73k\u503c\u3002\u800c\u8f6e\u5ed3\u7cfb\u6570\u5219\u901a\u8fc7\u8ba1\u7b97\u5404\u4e2a\u70b9\u4e0e\u5176\u81ea\u8eab\u805a\u7c7b\u7684\u7d27\u5bc6\u5ea6\u4ee5\u53ca\u4e0e\u90bb\u8fd1\u805a\u7c7b\u7684\u5206\u79bb\u5ea6\u6765\u8bc4\u4f30\u805a\u7c7b\u6548\u679c\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"Python\u8c03\u7528KMeans\u7684\u65b9\u6cd5\u5305\u62ec\u5bfc\u5165\u5fc5\u8981\u7684\u5e93\u3001\u51c6\u5907\u6570\u636e\u3001\u521d\u59cb\u5316KMeans\u6a21\u578b\u3001\u62df\u5408\u6a21\u578b\u3001\u83b7\u53d6\u805a\u7c7b\u7ed3\u679c\u3001 […]","protected":false},"author":3,"featured_media":936752,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[37],"tags":[],"acf":[],"_links":{"self":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/936751"}],"collection":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/comments?post=936751"}],"version-history":[{"count":"1","href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/936751\/revisions"}],"predecessor-version":[{"id":936753,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/936751\/revisions\/936753"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media\/936752"}],"wp:attachment":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media?parent=936751"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/categories?post=936751"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/tags?post=936751"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
![\"PYTHON\u5982\u4f55\u8c03\u7528kmeans\"](\"https:\/\/cdn-kb.worktile.com\/kb\/wp-content\/uploads\/2024\/04\/25072859\/b2af911a-80db-4838-98ad-66888bbe7b85.webp\")
<\/p>\n