Least squares method of estimation and removing of outliers

Hello everyone! I have i=1...1000 samples and the system is described by y(i)=a1*y(i-1)+a2*y(i-2)+b*u(i-1) where u is our input and y is output .How can i apply the least square method for estimate a1,a2,b and remove the outliers before? Thank you for your help

Answer 1

[Ajay Goyal]

Solve inverse of the problem and use stepwise linear regression (stepwiselm) or simply least square method (regress). Then generated equation can be inverted