本文共 11258 字，大约阅读时间需要 37 分钟。
主成分分析 独立成分分析
by Moshe Binieli
由Moshe Binieli
This article will explain you what Principal Component Analysis (PCA) is, why we need it and how we use it. I will try to make it as simple as possible while avoiding hard examples or words which can cause a headache.
本文将向您解释什么是主成分分析(PCA)，为什么需要它以及如何使用它。 我将尝试使其尽可能简单，同时避免使用可能引起头痛的硬示例或单词。
A moment of honesty: to fully understand this article, a basic understanding of some linear algebra and statistics is essential. Take a few minutes to review the following topics, if you need to, in order to make it easy to understand PCA:
诚实的时刻：要完全理解本文，对一些线性代数和统计量的基本理解是必不可少的。 如果需要，请花几分钟时间回顾以下主题，以使其易于理解PCA：
Let’s say we have 10 variables in our dataset and let’s assume that 3 variables capture 90% of the dataset, and 7 variables capture 10% of the dataset.
假设我们的数据集中有10个变量，并假设3个变量捕获了90％的数据集，而7个变量捕获了10％的数据集。
Let’s say we want to visualize 10 variables. Of course we cannot do that, we can visualize only maximum 3 variables (Maybe in future we will be able to).
假设我们要可视化10个变量。 当然我们不能这样做，我们只能可视化最多3个变量(也许将来我们可以)。
So we have a problem: we don’t know which of the variables captures the largest variability in our data. To solve this mystery, we’ll apply the PCA Algorithm. The output will tell us who are those variables. Sounds cool, doesn’t it? ?
所以我们有一个问题：我们不知道哪个变量捕获了数据中最大的可变性。 为了解决这个难题，我们将应用PCA算法。 输出将告诉我们那些变量是谁。 听起来很酷，不是吗？ ？
Our main goal is to figure out how many variables are the most important for us and stay only with them.
我们的主要目标是找出对我们来说最重要的变量有多少，并且只保留它们。
For this example, we will use the program “Spyder” for running python. We’ll also use a pretty cool dataset that is embedded inside “sklearn.datasets” which is called “load_iris”. You can read more about this dataset at .
对于此示例，我们将使用程序“ Spyder”运行python。 我们还将使用嵌入在“ sklearn.datasets”内部的非常酷的数据集，称为“ load_iris”。 您可以在了解有关此数据集的更多信息。
First of all, we will load the iris module and transform the dataset into a matrix. The dataset contains 4 variables with 150 examples. Hence, the dimensionality of our data matrix is: (150, 4).
首先，我们将加载虹膜模块并将数据集转换为矩阵。 数据集包含4个变量和150个示例。 因此，我们的数据矩阵的维数为：(150，4)。
import numpy as npimport pandas as pdfrom sklearn.datasets import load_iris
irisModule = load_iris()dataset = np.array(irisModule.data)
There are more rows in this dataset — as we said there are 150 rows, but we can see only 17 rows.
该数据集中有更多行-就像我们说的有150行，但我们只能看到17行。
The concept of PCA is to reduce the dimensionality of the matrix by finding the directions that captures most of the variability in our data matrix. Therefore, we’d like to find them.
PCA的概念是通过找到捕获我们数据矩阵中大多数可变性的方向来减少矩阵的维数。 因此，我们想找到它们。
It’s time to calculate the covariance matrix of our dataset, but what does this even mean? Why do we need to calculate the covariance matrix? How will it look?
现在是时候计算数据集的协方差矩阵了，但这到底意味着什么？ 为什么我们需要计算协方差矩阵？ 看起来如何？
is the expectation of the squared deviation of a random variable from its mean. Informally, it measures the spread of a set of numbers from their mean. The mathematical definition is:
是随机变量与其平均值的平方偏差的期望值。 非正式地， 它根据一组均值来衡量一组数字的传播。 数学定义为：
is a measure of the joint variability of two random variables. In other words, how any 2 features vary from each other. Using the covariance is very common when looking for patterns in data. The mathematical definition is:
是两个随机变量联合变量的度量。 换句话说，任何两个功能如何彼此不同。 在数据中查找模式时，通常使用协方差。 数学定义为：
From this definition, we can conclude that the covariance matrix will be symmetric. This is important because it means that its eigenvectors will be real and non-negative, which makes it easier for us (we dare you to claim that working with complex numbers is easier than real ones!)
根据这个定义，我们可以得出结论，协方差矩阵将是对称的。 这很重要，因为这意味着它的特征向量将是实数和非负数，这使我们更容易(我们敢于要求与实数相比，使用复数更容易！)
After calculating the covariance matrix it will look like this:
计算协方差矩阵后，它将如下所示：
As you can see, the main diagonal is written as V (variance) and the rest is written as C (covariance), why is that?
如您所见，主对角线写为V (方差)，其余对角线写为C (协方差)，为什么？
Because calculating the covariance of the same variable is basically calculating its variance (if you’re not sure why — please take a few minutes to understand what variance and covariance are).
因为计算同一个变量的协方差基本上就是计算其方差(如果不确定为什么，请花几分钟来了解什么是方差和协方差)。
Let’s calculate in Python the covariance matrix of the dataset using the following code:
让我们使用以下代码在Python中计算数据集的协方差矩阵：
covarianceMatrix = pd.DataFrame(data = np.cov(dataset, rowvar = False), columns = irisModule.feature_names, index = irisModule.feature_names)
We’re not interested in the main diagonal, because they are the variance of the same variable. Since we’re trying to find new patterns in the dataset, we’ll ignore the main diagonal.
我们对主对角线不感兴趣，因为它们是同一变量的方差。 由于我们正在尝试在数据集中查找新的模式，因此我们将忽略主对角线 。
Since the matrix is symmetric, covariance(a,b) = covariance(b,a), we will look only at the top values of the covariance matrix (above diagonal).
由于矩阵是对称的，所以协方差(a，b)=协方差(b，a)， 我们将仅查看协方差矩阵的最高值(对角线以上) 。
Something important to mention about covariance: if the covariance of variables
关于协方差重要的事情要提到：如果变量的协方差
a and b is positive, that means they vary in the same direction. If the covariance of a and b is negative, they vary in different directions.
a和b为正 ，表示它们沿相同方向变化。 如果a和b的协方差为负 ，则它们在不同的方向上变化。
As I mentioned at the beginning, eigenvalues and eigenvectors are the basic terms you must know in order to understand this step. Therefore, I won’t explain it, but will rather move to compute them.
正如我在开始时提到的那样，特征值和特征向量是您必须了解的基本术语，才能理解此步骤。 因此，我不会解释它，而是会进行计算。
The eigenvector associated with the largest eigenvalue indicates the direction in which the data has the most variance. Hence, using eigenvalues we will know what eigenvectors capture the most variability in our data.
与最大特征值关联的特征向量指示数据具有最大方差的方向。 因此，使用特征值，我们将知道哪些特征向量捕获了数据中最大的可变性。
eigenvalues, eigenvectors = np.linalg.eig(covarianceMatrix)
This is the vector of the eigenvalues, the first index at eigenvalues vector is associated with the first index at eigenvectors matrix.
这是特征值的向量，在特征值向量处的第一索引与在特征向量矩阵处的第一索引相关联。
The eigenvalues:
特征值：
The eigenvectors matrix:
特征向量矩阵：
The eigenvalues tells us the amount of variability in the direction of its corresponding eigenvector. Therefore, the eigenvector with the largest eigenvalue is the direction with most variability. We call this eigenvector the first principle component (or axis). From this logic, the eigenvector with the second largest eigenvalue will be called the second principal component, and so on.
特征值告诉我们在其相应特征向量方向上的变化量。 因此，特征值最大的特征向量是变化最大的方向。 我们称这个特征向量为第一主成分(或轴)。 根据此逻辑，具有第二大特征值的特征向量将被称为第二主成分，依此类推。
We see the following values:[4.224, 0.242, 0.078, 0.023]
我们看到以下值：[4.224、0.242、0.078、0.023]
Let’s translate those values to percentages and visualize them. We’ll take the percentage that each eigenvalue covers in the dataset.
让我们将这些值转换为百分比并将其可视化。 我们将获取每个特征值在数据集中所占的百分比。
totalSum = sum(eigenvalues)variablesExplained = [(i / totalSum) for i in sorted(eigenvalues, reverse = True)]
As you can clearly see the first and eigenvalue takes 92.5% and the second one takes 5.3%, and the third and forth don’t cover much data from the total dataset. Therefore we can easily decide to remain with only 2 variables, the first one and the second one.
正如您可以清楚地看到的那样， 第一个和特征值占92.5％ ， 第二个占5.3％，而第三个和第四个并没有覆盖整个数据集中的大量数据。 因此，我们可以轻松地决定只保留两个变量 ，第一个和第二个。
featureVector = eigenvectors[:,:2]
Let’s remove the third and fourth variables from the dataset. Important to say that at this point we lose some information. It is impossible to reduce dimensions without losing some information (under the assumption of general position). PCA algorithm tells us the right way to reduce dimensions while keeping the maximum amount of information regarding our data.
让我们从数据集中删除第三个和第四个变量。 重要的一点是，我们现在会丢失一些信息。 在不损失某些信息的情况下减小尺寸是不可能的(在一般情况下)。 PCA算法为我们提供了减少尺寸的正确方法，同时又保留了有关我们数据的最大信息量。
And the remaining data set looks like this:
其余数据集如下所示：
We want to build a new reduced dataset from the K chosen principle components.
我们想从K个选定的主成分构建一个新的简化数据集。
We’ll take the K chosen principles component (k=2 here) which gives us a matrix of size (4, 2), and we will take the original dataset which is a matrix of size (150, 4).
我们将选取K个选定的原则分量(此处k = 2)，这将为我们提供一个大小为(4，2)的矩阵，我们将为原始数据集提供一个大小为(150，4)的矩阵。
We’ll perform matrices multiplication in such a way:
我们将以以下方式执行矩阵乘法：
featureVectorTranspose = np.transpose(featureVector)datasetTranspose = np.transpose(dataset)newDatasetTranspose = np.matmul(featureVectorTranspose, datasetTranspose)newDataset = np.transpose(newDatasetTranspose)
After performing the matrices multiplication and transposing the result matrix, these are the values we get for the new data which contains only the K principal components we’ve chosen.
在执行矩阵乘法并转置结果矩阵之后，这些就是我们从新数据中获得的值，这些新数据仅包含我们选择的K个主成分。
As (we hope) you can now see, PCA is not that hard. We’ve managed to reduce the dimensions of the dataset pretty easily using Python.
正如我们现在所希望的那样，PCA并不难。 我们已经成功地使用Python轻松减少了数据集的尺寸。
In our data set, we did not cause serious impact because we removed only 2 variables out of 4. But let’s assume we have 200 variables in our data set, and we reduced from 200 variables to 3 variables — it’s already becoming more meaningful.
在我们的数据集中，我们没有造成严重影响，因为我们仅从4个变量中删除了2个变量。但是，假设我们的数据集中有200个变量，并且从200个变量减少到3个变量-它已经变得越来越有意义。
Hopefully, you’ve learned something new today. Feel free to contact or on Linkedin for any questions.
希望您今天学到了一些新知识。 如有任何疑问，请随时通过Linkedin与或联系。
翻译自:
主成分分析 独立成分分析
转载地址：http://ihrwd.baihongyu.com/