labels command
by Leon Unger
Hi there,
I tried to use the options of the labels command --to-file and
--from-file and get as response that GRETL does not know what to with it.
I'm using the windows snapshop.
Leon
12 years, 4 months
Using bfgs
by clarodina＠lycos.com
Allin Cottrell,
Make some adjust to the script but don't know does is there error
d1 d2 and d3 are different data series
Want regress d1 against d2 and d3 without const
The output would be a regress d1hat
Do a adf on d1 - d1hat
Use the bfgsmax to determine the coefficients of d2 and d3 maximising t from the adf of d1 - d11hat
function scalar adf_t (matrix *b, series y, series d2, series d3)
series d0 = y - (b[1]*d2 + b[2]*d3)
adf 0 d0 --c --quiet
scalar tau = $test
return tau
end function
Using the above script gives a error
The convergence criterion was not met
Error executing script: halting
12 years, 4 months
Optimization
by clarodina＠lycos.com
Hi PS,
Want to have the coefficients maximising t value
Matlab have a add on doing the maximisation
How to do the maximising on gretl?
Clarodina
12 years, 4 months
Adding one variable for Autoregressive Conditional Duration
by Josephine Sudiman
Dear Gretl Mailing List Users,
My apologise to email you at this time of year. First of all, I wish that you will have a Merry Christmas which will come in less than 48 hours. I am honestly very grateful to join this mailing list, it is always very supportive.
Once I wrote an email in GRETL mailing list to ask help on how to code for the Maximum Likelihood Estimation to estimate the ACD model on GRETL and I have the great help from Prof. Lucchetti in this issue.
Now, I am begging for assistance again. It is still about the ACD model, but my specification is:
EP_DUR(t) = x0 + x1 P_DUR(t-1) + x2 EP_TIME(t-1) + x3 SPREAD (t-1)
where:
EPTIME is the expected duration
PTIME is the actual duration (i got the value of this series)
EPTIME is the expected duration at lag1
SPREAD is the spread at lag1; spread is difference between bid and ask (i got the value of this series)
I intend to apply Weilbull WACD (1,1) and Exponential EACD (1,1) to obtain the estimation of X0, x1, x2 and x3.
May question is where should I add the variable (SPREAD) in this model. I attached the previous GRETL program that Prof. Lucchetti wrote for me last time and my simple data in gdt. I am really sorry to disturb you all during this festive break. Many sincere thanks for any help and attention....
Kind regards,
Josephine
------------------------------------------------------------------------------------------------------------------------------------------
12 years, 4 months
Using BFGSmax
by clarodina＠lycos.com
Allin Cottrell,
Tks for the reply
d1 is another data series and not the fit value from d1 = b[1] * d2 + b[2] * d3
Want to regress d1 against d2 and d3 and find a d1^hat approximating d1
Ols uses error minimization but want to use maximisation of t value from the adf of d1^hat
Using the script you provide, the original d1 data series is replace from the d1^hat
Clarodina
12 years, 4 months
Icon View Refresh
by Leon Unger
Hi there,
In therms of bug hunting I recently found out that the Icon View (in
GRETL for Windows) is not refreshed properly when doing this
<>hansl
nulldata 50
bundle foo
loop for i=1..240 -q
if i=2*int(i/2)
matrix even_M$i={}
sprintf mn "%d", i
string bn = "even_M" ~ mn
foo[bn]=even_M$i
delete even_M$i
else
matrix odd_M$i={}
sprintf mn "%d", i
string bn = "odd_M" ~ mn
foo[bn]=odd_M$i
delete odd_M$i
endif
endloop
<>hansl
The bunch of matrix objects seems to be still there although delete them
but closing and reopening the icon view solves 'the problem'.
Merry Christmas!
Leon
12 years, 4 months
Using BFGSmax
by clarodina＠lycos.com
Hi,
Is there some steps examples on how to use BFGSmax to determine the alpha and beta of d2 and d3
d1= alpha * d2 + beta * d3
with maximization of t value from adf 0 d1
Clarodina
12 years, 4 months
Re: [Gretl-users] Working with models
by Frederic Beister
Hi,
I already figured out how to use the gnuplot command for variables - the
question is how I can generate a plot similar to the plot that is
obtained through the model-graph-menu (model vs. real, over time) from
the console. I'll look into those $-variables as mentioned earlier.
On 22.12.2011 15:38, Ignacio Diaz-Emparanza wrote:
> El 22/12/11 13:53, Frederic Beister escribió:
>> Now I have a model stored as m1 - from the GUI I know that I can plot
>> several graphs such as residuals etc.. How do I do this with the CLI or
>> how can I "export" the model to my local machine so that I can use it there?
>> (...)
>>"gnuplot m1" or similar doesn't work either as m1 is not a
>> variable.
>>
>> (...)
>
> Gretl uses the external program 'gnuplot' for making the plots. In fact,
> there is also a 'gnuplot' command in gretl:
>
> gnuplot yvars xvars --output=filename.pdf
>
> this command uses variables as parameters, not models. In particular
> this will send the graph to a pdf file. Please look at the gnuplot
> command in the gretl help or in the manual (Gretl Command Reference) for
> more information about this command.
12 years, 4 months
