Assignment -
Suppose we want to perform a simple linear regression analysis. Go to the D2L and get 'weight.txt' file. It contains two variables: age and weight. We can use the following code to do regression analysis.
Proc Reg data = d;
model weight = age;
run;
Let Y be a column vector of "weight". Let X be a matrix with the first column being a vector of one and the second column being a vector of "age". Let I be the identity matrix and J be a square matrix with its elements being 1.
Recall that
estimate of β = (X'X)-1X'Y
SST (sum of squares of total) = Y'[I-(1/n)J]Y
SSE (sum of squares due to error) = Y'[I-X(X'X)-1X']Y
SSR = SST - SSE.
Print the estimated coefficients and SST, SSE, and SSR in the output window using Proc IML. Check if your values are the same with the output of the Proc Reg.