library(MASS)
x <-c(22,52,60,42,47,65)
y<-c(41, 49,69,55,60,62)
Y Vector:
Y <- matrix(c(41,49,69,55,60,62), nrow = 6,
ncol = 1)
Y
## [,1]
## [1,] 41
## [2,] 49
## [3,] 69
## [4,] 55
## [5,] 60
## [6,] 62
Design Matrix
X <- matrix(c(1,22,1,52,1,60,1,42,1,47, 1,65),
nrow = 6,
ncol = 2,
byrow = TRUE)
X
## [,1] [,2]
## [1,] 1 22
## [2,] 1 52
## [3,] 1 60
## [4,] 1 42
## [5,] 1 47
## [6,] 1 65
X^T * X
t(X)%*%X
## [,1] [,2]
## [1,] 6 288
## [2,] 288 14986
X^T * Y
t(X)%*%Y
## [,1]
## [1,] 336
## [2,] 16750
(X^T * X)^-1
fractions(solve(t(X)%*%X))
## [,1] [,2]
## [1,] 7493/3486 -24/581
## [2,] -24/581 1/1162
B_hat, the vector of coefficient estimates
fractions((solve(t(X)%*%X))%*%(t(X)%*%Y))
## [,1]
## [1,] 17608/581
## [2,] 311/581
Y_hat, the fitted value vector
fractions(X%*%(solve(t(X)%*%X))%*%(t(X)%*%Y))
## [,1]
## [1,] 24450/581
## [2,] 33780/581
## [3,] 36268/581
## [4,] 11719/222
## [5,] 32225/581
## [6,] 37823/581
e_mat, the vector of residuals
hMat<-X%*%solve(t(X)%*%X)%*%t(X)
fractions(hMat%*%Y)
## [,1]
## [1,] 24450/581
## [2,] 33780/581
## [3,] 36268/581
## [4,] 11719/222
## [5,] 32225/581
## [6,] 37823/581
Checking with Y-Yhat:
fractions(Y-(X%*%(solve(t(X)%*%X))%*%(t(X)%*%Y)))
## [,1]
## [1,] -629/581
## [2,] -5311/581
## [3,] 3821/581
## [4,] 1285/581
## [5,] 2635/581
## [6,] -1801/581
Why are these two different?
Part Two: Input the data into R and create a scatter plot with the regression line.
mod<-lm(y~x)
plot(x,y, pch=16)
abline(coefficients(mod))