Median of dependent variable for LAD regression in output?
by andreas.karlsson＠ltv.se
Hi,
For the Least Absolute Deviation (LAD) regression analysis, gretl gives the
mean and standard deviation of the dependent variable in the output.
However, since LAD estimates the median regression and not the mean
regression, I think it would be pertinent to also give the median of the
dependent variable in the output. Could this be implemented in gretl,
please?
Thanks for a really great software.
Best regards
Andreas Karlsson
17 years
some small bugs
by Talha Yalta
I found these little bugs using 1.6.2 under windows:
1- When trying to select several variables to display summary
statistics, I hold the ctrl button and start selecting individual
variables. Now, while pressing the left mouse button (or while having
it pressed) on a new variable, if I move the mouse pointer even by a
pixel, the whole previous selection is gone. The only way to select
multiple (not adjacent) variables is to make sure the mouse pointer
does not move while the left mouse button is pressed.
2- From the sample menu, selecting restrict based on criterion and
entering, for example, VALUE="" crashes the program (happens with "")
3- Pressing q (not Q) while renaming a model or graph closes the icon view.
4- Selecting "arrange icons" deletes the custom names given to models
returning back to Model1, Model2 etc. Similarly, model names are not
saved after saving a session so that I see Model1, Model2 etc.
However, double clicking a model, I can see the custom name in the
first row (so it was not completely lost).
Cheers,
Talha
--
"Money won't buy happiness, but it will pay the salaries of a large
research staff to study the problem." - Bill Vaughan
--
17 years
Ricardo G Silva/Serainternet/BR está ausente do escritório.
by ricardosilva＠serasa.com.br
Estarei ausente do escritório a partir de 05/04/2007 e não retornarei até
10/04/2007.
Estarei ausente do escritório até o dia 10/05/2006.
Em caso assuntos "urgentes" estarei recebendo Notes via Escritório Virtual.
Para assuntos referentes a dados de Inadimplência, favor entrar em contato
com Juliana (8868) ou Paulo Nogueira (9381).
17 years
Restrictions on cointegration vectors
by Riccardo (Jack) Lucchetti
Dear all,
Allin and I have just slightly modified the output of the "restrict"
command after estimating a VECM. You now get, together with the test
proper, the restricted estimate of the cointegration matrix \beta and of
the loadings \alpha. Please note that the restrictions are applied to all
cointegration vectors: restrictions on individual vectors are not handled
yet. This is planned, but for a longer timeframe.
The attached script reproduces two examples from Johansen (1995), OUP.
We'd like to have some feedback from you cointegration junkies out there
(Sven, are you there? ;-)). One of the things we're currently uncertain
on is: when the restriction are applied, should the restricted model
estimated in full? On one hand, this seems a bit overkill: especially if
the restrictions are rejected, it makes little sense to compute the
short-run parameters and all the rest. On the other hand, if we don't
estimate the model, you don't get a covariance matrix for the restricted
\betas, since you need the residuals for that. [footnote: It must be
said, though, that I'm not sure at the moment whether the formulae for
the asymptotic covariance matrix for \beta still hold under linear
restictions.].
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Facoltà di Economia "G. Fuà"
Ancona
17 years
Aboput the focus on the bootstrap and VECMs [Was: Re: [Gretl-users] some small bugs)]
by Andreas Karlsson
[snip]
> I agree that the session concept is not fully exploited at present,
> and that is something we will come back to and re-examine. Thanks
> for the suggestions. Right now we're focused on particular
> econometric functionality (the bootstrap and VECMs) but the UI does
> need reconsideration.
Regarding bootstrap, I would like to suggest looking in Chapter 11 of Cameron,
A. Colin & Trivedi, Pravin K. (2005), Microeconometrics: Methods and
Applications. Cambridge: Cambridge University Press. ISBN: 0-521-84805-9.
This book gives a very good and detailed description of using bootstrap in
econometrics, including cases with heteroscedastic errors, panel data, or GMM.
Best regards,
Andreas K.
17 years
Bivariate tests and test statistic calculator menus equal
by Andreas Karlsson
The menu entries Model > Bivariate tests... and Tools > Test statistic
calculator both have tests for difference of means and differences of
varainces, which give the same results. What is the idea behind having the same
tests in two different menus?
Regards,
Andreas K.
17 years
Ang. RE: [Gretl-users] Bootstrap news
by Andreas Karlsson
> Typically, the bootstrap sampling vector is likely to be much
> longer than the series in the current data set, so direct saving as
> a series is, in the general case, not possible. We could, however,
> offer to save the bootstrap series as a .gdt file in its own right,
> which you could then open and analyse.
It sounds good and useful.
Regards,
Andreas
17 years
RE: [Gretl-users] Bootstrap news
by Andreas Karlsson
I have tested the new bootstrap analysis for OLS. It is VERY good, and very
useful. Thanks a lot.
The only thing I am missing is the possibility to save the values of the samling
distribution as a new variable in the data set. This could be useful if one e.g.
want to plot a histogram of this distribution (the kernel density estimation
plot obtained from the bootstrap menu entry is nice, but I prefer histograms).
Further, if the bootstrap method could be implemented for the Least Absolute
Deviation regression method too, I would find this very useful.
Med vänliga hälsningar / Best regards
Andreas Karlsson
cottrell(a)wfu.edu skrev 2007-03-29 04:55:03 :
> Andreas K suggested a short while ago that gretl should offer
> various bootstrap options via the graphical interface (GUI). I
> agreed at the time that this was a good idea. Here's a note on
> progress so far (in CVS, and the current gretl snapshot for Windows).
>
> First, the limitation: all of the following new stuff applies only
> to single-equation models estimated via OLS. (Though please note
> that we already offer bootstrapped confidence intervals for impulse
> response functions in relation to VARs.)
>
> When you estimate a model via OLS in the GUI, the model viewer
> window has a menu bar, including items labeled "Analysis" and "Tests".
>
> New under "Analysis": there's a "Bootstrap..." item. This opens a
> dialog box where you get to make five choices:
>
> (1) The variable/coefficient to examine.
>
> (2) "Confidence interval" vs "Studentized confidence interval" vs
> "P-value".
>
> (3) "Resampled residuals" vs "Simulate normal errors".
>
> (4) Number of replications (default 1000).
>
> (5) Show graph of bootstrap sampling distribution (no/yes).
>
> Notes:
>
> In relation to (1): you only get to examine one coefficient at a
> time by this particular means.
>
> On (2): the default (95%) confidence interval is based directly on
> the quantiles of the bootstrap coefficient estimates. The
> "studentized" version is as per Davidson and MacKinnon's
> "Econometric Theory and Methods" (ETM), chapter 5: at each bootstrap
> replication a t-ratio is formed as (a) the difference between the
> current and the baseline coefficient estimate, divided by (b) the
> baseline estimated standard error. Then the confidence interval is
> formed based on the quantiles of this t-ratio, as explained in ETM.
> The "P-value" variant is, again, as explained in ETM.
>
> On (3): you get to choose between resampling with replacement of the
> original residuals (rescaled as suggested in ETM), and simulated
> normal errors with the empirically given variance. Andreas suggested
> including the option of "case resampling" (that is, resampling the
> (y, X) pairs rather than the residuals. I have not implemented this
> to date for two reasons: first, it seems statistically dodgy, and
> second it is considerably more burdensome from the computational
> viewpoint. (You can economize substantially if the X matrix is
> treated as constant across the bootstrap replications.)
>
> Point (4) should be mostly self-explanatory. However, when you're
> doing a (1 - alpha) confidence interval, then, as explained in ETM,
> it is desirable that alpha(B + 1)/2 is an integer (where B is the
> number of replications). So gretl adjusts the user-chosen B value
> to ensure this is the case.
>
> Point (5) again should be self-explanatory: you can get gretl to
> make a graph of the density of the bootstrapped coefficient or t-
> ratio. This option employs gretl's kernel density estimation facility.
>
> The above all pertains to the "Analysis/Bootstrap" menu item. In
> addition you have options under "Tests/Linear restrictions". The
> restrictions dialog now has a "Use bootstrap" check box. If you
> check this, you get a bootstrapped F-test for whatever set of linear
> restictions you have entered. The methodology is as described in
> ETM for bootstrapped P-values.
>
> Autoregressive models: If the set of regressors includes the first
> lag of the dependent variable this should be handled correctly: the
> bootstrap data sets are calculated recursively, taking into account
> the autoregression. Please note that higher-order autoregressions
> are _not_ currently recognized and handled appropriately.
>
> In script mode: For single-equation models estimated via OLS, you
> can append the --boot flag to the "restrict" command to get
> bootstrapped tests. You can also set the default number of
> bootstrap replications using the "set" command with "bootrep"
> parameter. For example:
>
> set bootrep 10000
>
> Testing and comments welcome!
>
> Allin.
> _______________________________________________
> Gretl-users mailing list
> Gretl-users(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-users
17 years
H-P and end point problem
by fe_jasa
Dear Colleagues
I want to correct the end point problem of H-P. I have done the
instruction "addobs 12", for quarterly data, an ARIMA (to Y) and
obtained the forecast values (Ye). Now I am in conditions of applying
the H-P filter. But I have no solution to put the values of Y and Ye in
one variable. I have tried "Yn = Y + Ye", and also some matrix
operations, without success...
What must I do to have a variable with the "n" values of Y and the "12"
final values of "Ye" ?
Best regards
--
><><><><><><><><><><><><><>
João Sousa Andrade
Faculdade de Economia
Av. Dias da Silva, 165
3004-512 Coimbra
Tel: 351 239 790535
Fax: 351 239 790514
http://www4.fe.uc.pt/jasa
><><><><><><><><><><><><><>
17 years
