# Covariance- v. correlation-matrix based PCA

In principal component analysis (PCA), one can choose either the covariance matrix or the correlation matrix to find the components. These give different results because, I suspect, the eigenvectors between both matrices are not equal. (Mathematically) similar matrices have the same eigenvalues, but not necessarily the same eigenvectors. Several questions: (1) Why this difference? (2) Does PCA make sense, if you can get two different answers? (3) Which of the two methods is ‘best’? (4) Since PCA operates on standardized (not) raw data in both cases, i.e., scaled by their standard deviation, does it make sense to use the results to draw conclusions about the dominance of variation for the actual, unstandardized data?

The problem with not standardizing, i.e. with not scaling the variables by their standard deviation, is that if, for example, one variable is measured in centimeters and another in dollars, then changing centimeters to meters can actually change the eigenvectors, so an arbitrary choice of units can alter the results. Hence I'd use the correlation matrix.

Answered by Michael Hardy on December 5, 2020

## Related Questions

### Integrate $int_{-infty}^{infty} frac{dx}{1+x^{12}}$using partial fractions

2  Asked on November 30, 2020

### Galois connection for annhilators

1  Asked on November 30, 2020

### How to compare the growth rate between $lnln n$ and $2^{lg^* n}$

0  Asked on November 30, 2020 by aesop

### How do I use only NAND operators to express OR, NOT, and AND?

0  Asked on November 29, 2020 by user831636

### Showing that sum of first $998$ cubes is divisible by $999$

2  Asked on November 29, 2020 by rebronja

### Choosing two points from $[0,1]$ probability

2  Asked on November 29, 2020 by ucei

### Growth of Limits of $n$-th Terms in Series

1  Asked on November 29, 2020

### Why does $f^{(n)}(x)=sin(x+frac{npi}{2})$ for $f(x)=sin(x)$?

1  Asked on November 29, 2020 by cxlim

### There are $6$ digits containing $1$ and $0$, only problem is that $0$’s can’t be next to each other

1  Asked on November 29, 2020 by yaz-alp-ersoy

### How to show closure of ball of radius r/2 is a subset of ball with radius r

2  Asked on November 29, 2020 by hi-im-epsilon

### Let $M$ be a non-empty set whose elements are sets. What are $F={A×{A} : A⊆M, A≠∅}$ and $⋃F$?

1  Asked on November 28, 2020 by andrea-burgio

### Determinant of a linear transform between two different vector spaces with the same dimension

2  Asked on November 28, 2020 by lyrin

### Prove $int_{mathbb{R}^d} frac{|e^{ilangle xi, y rangle} + e^{- ilangle xi, y rangle} – 2|^2 }{|y|^{d+2}}dy = c_d |xi|^2$

2  Asked on November 28, 2020 by nga-ntq

### How to make an algorithm to check if you have won on a Lotto?

0  Asked on November 28, 2020 by jaakko-seppl

### For every set exists another stronger set

0  Asked on November 28, 2020 by 45465

### Find a formula for the general term $a_n$ of the sequence, assuming that the pattern of the first few terms continues.

1  Asked on November 27, 2020 by andrew-lewis

### Examples of irreducible holomorphic function in more than one variable.

1  Asked on November 27, 2020 by alain-ngalani

### Given $log_2(log_3x)=log_3(log_4y)=log_4(log_2z)$, find $x+y+z$.

3  Asked on November 27, 2020 by hongji-zhu

### Which of the following statements is correct?

1  Asked on November 27, 2020 by user469754