Other Considerations for Matrix-Vector Equations

Linear Algebra for Data Science in R

Eric Eager

Data Scientist at Pro Football Focus

More Equations than Unknowns

Row Reduction (By Hand, Difficult for Big Problems)
Least Squares (If More Rows Than Columns - Used in Linear Regression)
Singular Value Decomposition (If More Columns Than Rows - Used in Principal Component Analysis)
Generalized or Pseudo-Inverse

library(MASS)
print(A)

     [,1] [,2]
[1,]    2    3
[2,]   -1    4
[3,]    1    7

ginv(A)

          [,1]        [,2]       [,3]
[1,] 0.3333333 -0.30303030 0.03030303
[2,] 0.0000000  0.09090909 0.09090909

ginv(A)%*%A

     [,1]          [,2]
[1,]    1 -1.110223e-16
[2,]    0  1.000000e+00

A%*%ginv(A)

           [,1]       [,2]      [,3]
[1,]  0.6666667 -0.3333333 0.3333333
[2,] -0.3333333  0.6666667 0.3333333
[3,]  0.3333333  0.3333333 0.6666667

print(A)

     [,1] [,2]
[1,]    2    3
[2,]   -1    4
[3,]    1    7

print(b)

1 7 8

x <- ginv(A)%*%b
A%*%x

     [,1]
[1,]    1
[2,]    7
[3,]    8

Linear Algebra for Data Science in R