Regression

Non parametric regression using kernels to estimate density function of residuals.

It will first to a simple regression using least squares to get a good start value. Then it will use that start value for the non parametric regression where we maximize the likelihood based on non parametric methods to estimate the probability density function.

Just type '[B]=KLMHR2(X,y)' and it will work for you.

( Where X is your matrix where each column is a explanatory variable and y is a vector which you are trying to predict )