On Mon, 23 Apr 2012, Henrique Andrade wrote: True, you can't save models by name inside a function. This is part of the "encapsulation" idea for gretl functions: their ability to produce "side effects" is strictly circumscribed. Basically, functions are allowed to print stuff and/or return a value, and nothing else. But if you want to return multiple objects (in a user-friendly form) you can use the bundle data-type for that purpose. (See also http://www.wfu.edu/~cottrell/tmp/extending.pdf ) I'm not seeing a problem with that here. This works OK: <hansl> function void testarima (series y) loop p=1..2 loop q=1..2 arima p 1 q ; y --nc fcast Y_hat_$p_1_$q print Y_hat_$p_1_$q endloop endloop end function open fedstl.bin data exbzus dataset addobs 12 testarima(exbzus) </hansl> Of course the series generated by "fcast" are local to the function -- they won't be available outside the function unless you return then in some way. Here are some comments on your function: 1. Using "set halt_on_error off" is a bad idea (perhaps we should get rid of it): if you do that then any errors will cascade and error messages will likely be hard to understand. Do use, instead, the "catch" modifier, and check the $error variable afterwards. This is important for functions that invoke commands such as arima, where failure of convergence is a live possibility. 2. There's no need to "set echo off" or "set messages off" inside a function; that's the default behavior. But you can set those things on for debugging. 3. It's good gretl programming style to use the string representation of loop indices only where a string is actually needed; otherwise just use the numerical value of the index. So: fcast Y_hat_$P_1_$Q # OK, strings needed arima P 1 Q # strings not needed, don't use "arima $P 1 $Q" 4. Consider limiting the sum of the AR and MA orders to something sane. Your function as written allows up to arima(12, 1, 12), which would surely be grossly over-parameterized. 5. There seem to be some problems with initialization of your arima models that include dummies for observations where the plain arima residuals are greater than two standard deviations. We use nonlinear least squares for initialization of arima with exogenous regressors: that works quite well in some cases but apparently does not work well here. You might consider using "set initvals" to specify your own initialization. For example, in the arima(1,1,1) case: <hansl> diff y ols d_y d_y(-1) x1 x2 matrix m = {$coeff[1], 0.001, $coeff[2], $coeff[3]} set initvals m arima 1 1 1 ; y x1 x2 --nc </hansl> Allin Cottrell

on Wed, 25 Apr 2012, Henrique Andrade wrote: What do you want to do here, exactly? "teste_t" is a matrix, so its absolute value is a matrix too. How can you base an "if" condition on a multiple-valued object? -------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti --------------------------------------------------

Em 25 de abril de 2012, Allin escreveu: On Thu, 26 Apr 2012, Riccardo (Jack) Lucchetti wrote: Dear Jack and Allin, Sorry if I wasn't clear enough on my messages. Blame on my poor English ;-) Let me try to explain what is the main goal of the AutoARIMA function: 1. The user selects a time series; 2. Determines the max number os AR(p) and MA(q) terms; 3. The AutoARIMA function will estimate all the possible combinations of ARIMA(p,1,q), 4. Every time the estimated model exhibits its roots outside the unit circle and all its coefficients have statistical significance, AutoARIMA will save the predictions of that model. Answering the other question that Allin did... Em 25 de março de 32012, Allin escreveu: On Wed, 25 Apr 2012, Henrique Andrade wrote: Sorry again if I wasn't clear enough :-( I was talking about the attached function (and not the one in the body of the e-mail message). To avoid more mistakes I'm attaching it again. Kind regards, Henrique Andrade* *

Dear Allin, Thanks a lot for your valuable advices and fixing. Now my function is working :-) Best regards, Henrique 2012/4/26 Allin Cottrell <cottrell(a)wfu.edu&gt; -- Henrique Andrade* *

On Wed, 9 May 2012, Anutechia Asongu wrote: There are no such examples in the practice files. However, there are examples of use of the mle command in the Gretl User's Guide and I presume that use of mle would be the preferred way of estimating a threshold model. You just need the loglikelihood function for the model you want to estimate. Allin Cottrell

Thanks Sven S ________________________________ From: Sven Schreiber <svetosch(a)gmx.net&gt; To: gretl-users(a)lists.wfu.edu Sent: Thursday, May 10, 2012 9:48 AM Subject: Re: [Gretl-users] Threshold regressions in Practice files On 05/09/2012 11:15 PM, Allin Cottrell wrote: Perhaps it's even simpler than that, depending on what's meant with threshold regression here: If the threshold is known, you can just create a dummy variable defined as (x>threshold)  and interact that dummy with x. -sven _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users

Thanks Allin ________________________________ From: Allin Cottrell <cottrell(a)wfu.edu&gt; To: Gretl list <gretl-users(a)lists.wfu.edu&gt; Sent: Thursday, May 10, 2012 1:48 PM Subject: Re: [Gretl-users] Threshold regressions in Practice files On Thu, 10 May 2012, Sven Schreiber wrote: Footnote: I see that Bruce Hansen has R code available for the sort of threshold regression he estimated in his 2000 Econometrica paper (where the threshold is unknown in advance) http://www.ssc.wisc.edu/~bhansen/progs/ecnmt_00.html That could be run from gretl, or the R code could be translated to hansl. (Hansen uses least squares; it turns out that ML is quite awkward for this sort of model.) Allin Cottrell _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users