Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Model Selection and Choices

From   Paul Millar <>
Subject   Re: st: Model Selection and Choices
Date   Wed, 27 Apr 2011 12:41:46 -0400

Usually the BIC statistic (or alternately the AIC) is used for model
selection when the model uses maximum likelihood, using the
postestimation -estat(ic)- command.  Another method is the
Proportional Reduction in Errors (PRE) method (i.e. the model that has
the fewest errors in prediction).  There is a postestimation command
for this -pre- available from ssc, however, I am not sure I have
tested it for the type of time series models you are using.  -pre- has
the advantage of working for any model type.  If it doesn't seem to
work for the models you are using, send me the commands and a data
sample and I will get it working for you.

- Paul

On Wed, Apr 27, 2011 at 1:47 AM, Muhammad Anees <> wrote:
> Hi All!
> I have a straight question on time series modelling. I have a 37 years
> annualized time series data on five variables. I want to check if
> linear or non linear modelling would be best in my case. Is there any
> way using stata to obtain results and to select model that best
> reflects my specific time series data.
> --
> Muhammad Anees
> MSc in Economics
> The University of Sheffield
> United Kingdom
> *
> *   For searches and help try:
> *
> *
> *

- Paul Millar
Criminal Justice and Legal Studies
Nipissing University
North Bay, Ontario, Canada

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index