如何在PCA期间保留类标签

Question

在执行PCA时如何保留标签？我看了2个教程，他们完全忽略了这一点：教程

这是我的代码：

combinedOutputDataFrame = pd.DataFrame(resultArray)
# Separating out the features
x = combinedOutputDataFrame.loc[:, 0:31].values
# Separating out the target
y = combinedOutputDataFrame.loc[:,[32]].values
from sklearn.decomposition import PCA
pca = PCA(n_components=3)
principalComponents = pca.fit_transform(x)
principalDataFrame = pd.DataFrame(data = principalComponents
, columns = ['principal component 1', 'principal component 2', 'principal component 3'])
finalDf = pd.concat([principalDataFrame, combinedOutputDataFrame[[32]]], axis = 1)

不过，我如何确定主要组件的顺序？

principalComponents
array([[129.58602603, -21.59786631,  -6.84613849],
[-39.42963482,  35.19985695,  19.86945922],
[ 54.81949577,  -5.96905719, -76.57776259],
...,
[ 69.21840475, -35.17983093, -39.66853653],
[ 18.91508026, -41.64341368,   0.21503516],
[145.91595004, 127.82236242, 115.14571367]])

我的最终目标是将其可视化，并用相应的类为图上的每个点上色。但是，在执行PCA后，我如何在数据上添加标签？

Answer 1

组件已经按降序排列，从解释差异最大的组件到解释差异最小的组件。你可以用pca.explained_variance_ratio_打印出解释的方差比来检查这一点

import numpy as np
from sklearn.decomposition import PCA
# just a random matrix 
rand_matrix = np.random.rand(30,6)
pca = PCA(n_components=3)
principalComponents = pca.fit_transform(rand_matrix)
print(pca.explained_variance_ratio_)
Out: 
array([0.28898895, 0.22460396, 0.16874681])