T-SNE降维可视化聚类结果（二维可视化、三维可视化）

一、二维可视化

1.1 生成数据

import pandas as pd
import numpy as np
df = {"A":np.random.normal(0, 1, 1000),
      "B":np.random.normal(0, 1, 1000),
      "C":np.random.normal(0, 1, 1000),
      "D":np.random.normal(0, 1, 1000)}
df = pd.DataFrame(df)
df.head()

输出：

	A	B	C	D
0	-1.197896	2.012483	-3.483592	-1.871044
1	-0.107903	-0.176872	-0.530986	0.308957
2	0.347445	-0.600894	0.434848	-0.496437
3	-1.303541	0.194091	-0.699378	-0.929843
4	1.144723	-0.310182	0.405988	1.393773

1.2 K均值聚类

from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=4)
k_fit = kmeans.fit(df)
predictions = k_fit.labels_
df['Clusters'] = predictions 
df.head()

输出：

A	B	C	D	Clusters
0	-1.197896	2.012483	-3.483592	-1.871044	2
1	-0.107903	-0.176872	-0.530986	0.308957	1
2	0.347445	-0.600894	0.434848	-0.496437	0
3	-1.303541	0.194091	-0.699378	-0.929843	1
4	1.144723	-0.310182	0.405988	1.393773	3

1.3 T-SNE

from sklearn.manifold import TSNE

Xtsne = TSNE(n_components=2).fit_transform(df)

import matplotlib.pyplot as plt
import seaborn as sns
dftsne = pd.DataFrame(data=Xtsne , columns = ['x1','x2'])
dftsne['cluster'] = predictions
plt.figure(figsize=(10,10))
sns.scatterplot(data=dftsne,x='x1',y='x2',hue='cluster',legend="full", palette='tab10', alpha=0.8)
plt.title('Clusters Visualized on TSNE 2D')
plt.show()

二、三维可视化

2.1 生成数据

import pandas as pd
import numpy as np
df = {"A":np.random.normal(0, 1, 1000),
      "B":np.random.normal(0, 1, 1000),
      "C":np.random.normal(0, 1, 1000),
      "D":np.random.normal(0, 1, 1000)}
df = pd.DataFrame(df)
df.head()

输出：

	A	B	C	D
0	0.313137	0.834052	1.089202	-0.761206
1	-0.245337	-0.395838	0.772860	0.976120
2	-1.163416	1.389154	0.643586	0.975716
3	0.162503	-0.413536	-1.052316	2.248309
4	-0.126101	0.196020	-0.634920	0.633622

2.2 K均值聚类

from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=4)
k_fit = kmeans.fit(df)
predictions = k_fit.labels_
df['Clusters'] = predictions 
df.head()

输出：

from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=4)
k_fit = kmeans.fit(df)
predictions = k_fit.labels_
df['Clusters'] = predictions 
df.head()

2.3 T-SNE

from sklearn.manifold import TSNE

Xtsne = TSNE(n_components=3).fit_transform(df)

from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
dftsne = pd.DataFrame(data=Xtsne , columns = ['x1','x2','x3'])
dftsne['cluster'] = predictions
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(111, projection='3d')
x = dftsne['x1']
y = dftsne['x2']
z = dftsne['x3']
ax.scatter(x, y, z,c = predictions, cmap="jet", marker="o")
plt.title("KMeans (N = 4)");
ax.set_xlabel('x1')
ax.set_ylabel('x2')
ax.set_zlabel('x3')
plt.show()