mixing matrices and series
by artur bala
Hello,
This is just for discussion. I used to think that in doing mathematical
calculations, the matrices need not to be mixed up with series. For the
sake of clarityone can not, let's say multiply a series by a matrix. Is
that right?
Well, in the following script this kind of multiplication works fine. Is
that normal -- I mean, an intentionnally built feature -- or should we
avoid such a mixed operation?
nulldata 100
series x1 = normal()
matrix unit = ones($nobs,1)
scalar prod = unit'*x1 # matrix * series
matrix mx1 = x1
scalar prod = unit'*mx1 # matrix * matrix
best,
artur
13 years
replace with
by alexkakashi＠libero.it
Hi,
I have the following problem. Let x(t) a time series with T=200 observations.
I must create a new time series y(t), with 250 observations, resampling with
replacement the elements of x. I have used the command
genr y = replace(x)
but gretl creates y(t) with 200 observations. Can I create y with 250
observations?
Thanks.
Best regard
Alessandro Attanasio
13 years
deleting matrices
by artur bala
As long as the matrices and the series share the same working space, can
we manage to delete a couple of matrices (or more) in a one run like we
do for series e.g.
delete series1 series 2 # this command does work
delete matrix1 matrix2 # this command does not
Best,
artur
13 years
Bug-hunting: GUI close button (Windows)
by Leon Unger
Hi there,
I found a bug in the GUI windows for e.g. MLE or GMM. There is the
typical Windows close button, but one can only close the window by using
cancel.
I've got a question concerning the script editor:
After replacing letters then the cursor jumps to the first line of the
script.
It is possible to implement that the cursor remembers the starting position?
Cheers
Leon
13 years
moran function
by Giuseppe Vittucci
I have written a little function to compute the Moran's I statistic of
spatial autocorrelation in gretl and uploaded it on the server (or I
tried to do so...I don't know if it was successful ;-)), but the related
script refers to data which cannot be found on the gretl server cause
there were no db suitable for spatial analysis.
I attach the files here for convenience.
They are taken from the package written by Pisati for stata (and are
publicly available): spatreg.
(This is the code you have to run in stata:
spatwmat using columbusswm.dta, name(W) standardize
use columbusdata.dta", replace
spatgsa crime, weights(W) moran
to reproduce the results of the script attached to my function)
I compared the results of my function with those produced by R (almost
the same) and stata (Pisati's code).
The latter are slightly different as far as the computation of the
standard deviation is concerned but I carefully checked the formulas in
the original papers (there are also references in the function) and they
should be right.
I have recently started working with spatial econometrics and I would
like to somehow try to integrate these models and statistics in my
preferred econometric software ;-)
This was just a first try...
See you
Giuseppe Vittucci
13 years
test for heteroskedacity in the tobit model
by Mari Zeyringer
Dear Gretl Users,
I would like to test for heteroskedacity in the tobit model. I have been
reading about it but could not find out what tests are available in Gretl
for the Tobit model.
I would very much appreciate any kind of help.
Thank you and all the best,
Marianne
13 years
