Changes in Version 1.1-0
o Fixed diagonal matrix computations in gstslshet(). Due to incorrect
trace computations, prior versions could lead to erroneous results.
However, for the Boston data in Piras (2009) - see the vignette for more
details - only slight differences occur.
Details:
The old command line
R> gammas<-as(Diagonal(,as.vector(gamma)),"sparseMatrix")
in the function gstslshet() was producing incorrect results in the computation of the following trace term (for R versions newer than 2.9.0):
R> tr1<-sum(diag(A1A1t%*%gammas%*%A1A1t%*%gammas))/2
For the Boston data results are only slightly different from those in Piras (2009)
(see sphet vignette for greater details).
With different data, differences could be quite large.
o added the zero.policy option as in spdep
o impacts() calculation for objects of class "sphet"
Changes in Version 1.2-0
o Added the function spreg to match stata and GeoDa.
Changes in Version 1.4-0
o Read the instrument and endogenous variable from the data
Changes in Version 1.4-1
o fix the residuals for model without intercept.
o put options for minimizations under control = list()
o solve a problem with the HAC. In particular there are now four options (see help).
Changes in Version 1.4-2
o fix a serious of issues with the instruments.
o endog and instruments are now either NULL or a formula type of object
Changes in Version 1.4-3
o intImpacts was exported from spdep on revision 530 on R-forge. Therefore, there are no more warning
when trying to access that function.
o loading much faster due to changes in spdep
o Replaced fields::rdist.earth with sp::spDists, which avoids importing from fields and provide a more accurate computation of the distance. In fact,
rdist.earth used a spheroid and spDists an ellipsoid (WGS84) for Great Circle distances.
o loading much faster due to changes in spdep