Dear Listers,
I'm working on the following problem of spatial autocorrelation model using Maximum likelihood function.
y0 = x*b0 + e0
y0= rho*W*y0 + x*b1 + e1
W is a spatial weight matrix. y0 is a dependent variable, x is an independent variable and rho is a correlation coefficient.
log likelihood function is = -n/2*ln(pi) -(n/2)*ln(1/n)(e0 - rho*e1)'(e0-rho*e1) + ln(I -rho*W)
where I is an identity matrix and W is a spatial weight matrix.
My question is:
how to present I - rho*W in the function above? rho should be a problem of this function but I have W in my data and wondering how to recall it to Mata.
Anyone has an idea of how to implement this in Mata language?
Many thanks in advance.
Kind regards,
Sungwook
I'm working on the following problem of spatial autocorrelation model using Maximum likelihood function.
y0 = x*b0 + e0
y0= rho*W*y0 + x*b1 + e1
W is a spatial weight matrix. y0 is a dependent variable, x is an independent variable and rho is a correlation coefficient.
log likelihood function is = -n/2*ln(pi) -(n/2)*ln(1/n)(e0 - rho*e1)'(e0-rho*e1) + ln(I -rho*W)
where I is an identity matrix and W is a spatial weight matrix.
My question is:
how to present I - rho*W in the function above? rho should be a problem of this function but I have W in my data and wondering how to recall it to Mata.
Anyone has an idea of how to implement this in Mata language?
Many thanks in advance.
Kind regards,
Sungwook
