1.算法思想


k-平均聚类的目的是：把n个点划分到k个聚类中，使得每个点都属于离他最近的均值（即聚类中心）对应的聚类。

2.聚类算法和分类算法

分类：根据文本的特征或属性，划分到已有的类别中。也就是说，这些类别是已知的，通过对已知分类的数据进行训练和学习，找到这些不同类的特征，再对未分类的数据进行分类。
聚类：压根不知道数据会分为几类，通过聚类分析将数据或者说用户聚合成几个群体，那就是聚类了。聚类不需要对数据进行训练和学习。就像小时候分校区，你家这块离哪个小学近，就去那个小学上学。

3.监督学习和非监督学习

image.png
在分类算法中，标签就是类别；而在回归算法中，标签就是具体的y值。

4.算法步骤

image.png

5.sklearn代码

  1. # coding:utf-8
  2. """
  3. ====================================
  4. Demonstration of k-means assumptions
  5. ====================================
  7. This example is meant to illustrate situations where k-means will produce
  8. unintuitive and possibly unexpected clusters. In the first three plots, the
  9. input data does not conform to some implicit assumption that k-means makes and
  10. undesirable clusters are produced as a result. In the last plot, k-means
  11. returns intuitive clusters despite unevenly sized blobs.
  12. """
  13. print(__doc__)
  15. # Author: Phil Roth <mr.phil.roth@gmail.com>
  16. # License: BSD 3 clause
  18. import numpy as np
  19. import matplotlib.pyplot as plt
  21. from sklearn.cluster import KMeans
  22. from sklearn.datasets import make_blobs
  24. plt.figure(figsize=(12, 12))
  26. n_samples = 1500
  27. random_state = 170
  28. X, y = make_blobs(n_samples=n_samples, random_state=random_state)  # 为聚类产生数据集
  30. # Incorrect number of clusters
  31. y_pred = KMeans(n_clusters=2, random_state=random_state).fit_predict(X)
  33. plt.subplot(221)
  34. plt.scatter(X[:, 0], X[:, 1], c=y_pred)
  35. plt.title("Incorrect Number of Blobs")
  37. # Anisotropicly distributed data
  38. transformation = [[0.60834549, -0.63667341], [-0.40887718, 0.85253229]]
  39. X_aniso = np.dot(X, transformation)
  40. y_pred = KMeans(n_clusters=3, random_state=random_state).fit_predict(X_aniso)
  42. plt.subplot(222)
  43. plt.scatter(X_aniso[:, 0], X_aniso[:, 1], c=y_pred)
  44. plt.title("Anisotropicly Distributed Blobs")
  46. # Different variance
  47. X_varied, y_varied = make_blobs(n_samples=n_samples,
  48.                                 cluster_std=[1.0, 2.5, 0.5],
  49.                                 random_state=random_state)
  50. y_pred = KMeans(n_clusters=3, random_state=random_state).fit_predict(X_varied)
  52. plt.subplot(223)
  53. plt.scatter(X_varied[:, 0], X_varied[:, 1], c=y_pred)
  54. plt.title("Unequal Variance")
  56. # Unevenly sized blobs
  57. X_filtered = np.vstack((X[y == 0][:500], X[y == 1][:100], X[y == 2][:10]))
  58. y_pred = KMeans(n_clusters=3,
  59.                 random_state=random_state).fit_predict(X_filtered)
  61. plt.subplot(224)
  62. plt.scatter(X_filtered[:, 0], X_filtered[:, 1], c=y_pred)
  63. plt.title("Unevenly Sized Blobs")
  65. plt.show()

最后跑出来的图
image.png
由于聚类算法目前在业务中需要了解的并不多，目前对这些代码暂时不深究，后续有需求再来完善本部分。