Re: Eureqa - Slightly off topic

[Michael Schmidt]

Michael, I'm not sure I completely understood. Are you saying that you would prefer some explanatory variables to appear in a model more than others (perhaps based on their variance)? One approach might be to start with allowing all variables, see which are used, and then remove the least desirable one, update the search, and repeat. This would be a way to iteratively guide the search toward what you want, but still have a data-driven result.

On Wed, Apr 24, 2013 at 6:16 AM, Michael D <michael...@gmail.com> wrote:

Now I don’t use Eureqa for my research, mostly because the models I am using currently seem to me to be too difficult for me to instantiate in Eureqa (but not in Excel!)… But I watch the discussions.

 For my modelling, I use AICc to help choose between forms of models.  In one experiment, I have, for instance, 14 experimental conditions in which I vary the temporal characteristics of inputs (to pigeons) that change over time, and I measure how their choices change over times.  The basic data are very stable (and very extensive) as I expose each of the birds to around 85 hours (sessions) of a single manipulation, and each session has up to 80 presentations of the temporal changes each over, say, a minute.  I may have 8 models.

 My problem is this – empirically, some of the manipulations result in more changes in choice over time (the behavior is controlled by the manipulation; more variance to be accounted for), and some less (less total variance).  AICc points to various models, but here’s my question:  Surely I should put more weight on results from conditions with greater variance?  But I have not seen a discussion of such things, and wonder whether any of you know of such a discussion. Maybe Burnham and Anderson discussed this, but I can’t check because (as is common), a student ran off with my copy…

 Any help anyone can give would be much appreciated.  But please remember, I’m not a mathematician, but a mere psychologist.

--
You received this message because you are subscribed to the Google Groups "Eureqa Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eureqa-group...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Dave Nunez]

"but a mere psychologist"  I get so angry when people say stuff like this, and sorry for the delay, I took psy 101 and found it so interesting back in the days, you are studying  complex systems, very non-linear with multiple variables to say the least and looking for correlations in data that is a very rigorous process as the likes that we go through.. don't be shy, cog dis it's all good, we deal with it every day.  -d

--
When things get too complicated, it sometimes makes sense to stop and wonder: Have I asked the right question?
-Enrico Bombieri

[Dave Nunez]

Hey Michael, no worries.. and sure, please send me the data that you have.  I can't promise you anything though.

Cheers, -d

On Mon, May 20, 2013 at 6:24 AM, Michael D <michael...@gmail.com> wrote:

Dave - thanks for that!  Actually, for years, I have been telling my UG and G students that if they are doing Psych as an easy option, they need to know that it is considerably more complex than the mature sciences; and because it is an immature and complex science, it's much more interesting and has many more creative possibilities than the mature sciences.  A few hear this.

 Perhaps, if you're interested, I could privately send you some papers and data and graphs, and see what you think?

 m