Hart:Stats: Difference between revisions

From OpenWetWare
Jump to navigationJump to search
No edit summary
No edit summary
Line 1: Line 1:
{{Template:Hart}}
{{Template:Hart}}
==Computing Interests==
==Computing Interests==
I recently wrote an agent based model for a three species metacommunity. This code is associated with my talk given at the most recent ESA.  The code has been written to run on a Linux node at the Vermont advanced computing center (VACC), but I have visualization code for the output and code that can visualize model runs.  Please contact me if you are interested.  You can see the model code [[Media: ESA_ABM.r | here]]
My main computing interests are in agent based models and Bayesian hierarchical models.  I program almost exclusively in R, although I do occasionally work with MatLab and Mathematica.


I am also interested in other computational methods as they apply to ecology such as evolutionary computation and cellular automata models, but I don't have much experience at the moment.  But I do have a simple GA written in R to solve the simple [http://www.duke.edu/vertices/update/win95/genalg.html one max] problem.  Here is my [http://www.uvm.edu/~ehart1/GA_Max_1.R solution]
Below are some code examples from manuscripts, talks and just my own edification.


Another project I am currently working on uses NetLogo to create individual based simulations using a simple rule set to explore metacommunity hypotheses.  You can explore with
Here is code for an agent based model of a three species metacommunity. The code has been written to run on a Linux node at the Vermont advanced computing center (VACC), but I have visualization code for the output and code that can visualize model runs.  Please contact me if you are interested.  You can see the model code [[Media: ESA_ABM.r | here]]
here [http://ccl.northwestern.edu/netlogo/ NetLogo] and you can see one of my original prototype models.


==Bayesian Methods==
The above model is a fancier version of an initial mockup I did in [http://ccl.northwestern.edu/netlogo/ NetLogo], you can see my code [[Media:Competition_colonization_tradeoff.nlogo| here]
I am interested Bayesian methods and am continually teaching myself.  A few blogs that I read regularly are those by [http://learnbayes.blogspot.com/ Jim Albert], and [http://www.stat.columbia.edu/~gelman/blog/ Andrew Gelman].  
 
I am also interested in other computational methods as they apply to ecology such as evolutionary computation and cellular automata models.  Here is a simple GA written in R to solve the simple [http://www.duke.edu/vertices/update/win95/genalg.html one max] problem.  Here is my [http://www.uvm.edu/~ehart1/GA_Max_1.R solution]
 
I do much of my Bayesian computation using WinBUGS, but I also hand code samplers when I need to.  Here are two examples.
 
-A home grown metropolis sampler [http://www.uvm.edu/~ehart1/MH_Norm.R MH_Norm:] fort estimating the parameters for a normal distribution from a vector of random normal numbers and then an example with linear regression
 
-Sample code from my hierarchical time series models, this is an example that runs on the VACC again for [[Media:VACC_Model_Runs.r|Chironomidae]]
 
==Bayesian Resources==
A few blogs that I read regularly are those by [http://learnbayes.blogspot.com/ Jim Albert], and [http://www.stat.columbia.edu/~gelman/blog/ Andrew Gelman].  


I also have used the following books to help teach myself
I also have used the following books to help teach myself
Line 24: Line 33:


Of course I also rely on my adviser [http://www.uvm.edu/~ngotelli/home.html Nick Gotelli's] book [http://www.amazon.com/Primer-Ecological-Statistics-Nicholas-Gotelli/dp/0878932690/ref=sr_1_2?ie=UTF8&s=books&qid=1219779461&sr=1-2 A primer of Ecological Statistics]
Of course I also rely on my adviser [http://www.uvm.edu/~ngotelli/home.html Nick Gotelli's] book [http://www.amazon.com/Primer-Ecological-Statistics-Nicholas-Gotelli/dp/0878932690/ref=sr_1_2?ie=UTF8&s=books&qid=1219779461&sr=1-2 A primer of Ecological Statistics]
==Software and Code==
I am also a died in the wool believer in the [http://cran.r-project.org/ power of R] and using it with [http://www.mrc-bsu.cam.ac.uk/bugs/ WinBUGS].
I have some original code that could be used by anyone, but is probably best used as a learning example.
[http://www.uvm.edu/~ehart1/MH_Norm.R MH_Norm:]  This has two parts, one is just estimating the parameters for a normal distribution from a vector of random normal numbers.  The next is a simple linear regression.  It uses a Metropolis-Hastings algorithm for this.  Its not the best documented code though, so I apologize.
I think one of the best editors for R in Windows in [http://www.sciviews.org/Tinn-R/ TINN-R].

Revision as of 13:50, 16 March 2010

Home        Papers        Teaching        Statistics and Computing        Talks        Research        CV       


Computing Interests

My main computing interests are in agent based models and Bayesian hierarchical models. I program almost exclusively in R, although I do occasionally work with MatLab and Mathematica.

Below are some code examples from manuscripts, talks and just my own edification.

Here is code for an agent based model of a three species metacommunity. The code has been written to run on a Linux node at the Vermont advanced computing center (VACC), but I have visualization code for the output and code that can visualize model runs. Please contact me if you are interested. You can see the model code here

The above model is a fancier version of an initial mockup I did in NetLogo, you can see my code [[Media:Competition_colonization_tradeoff.nlogo| here]

I am also interested in other computational methods as they apply to ecology such as evolutionary computation and cellular automata models. Here is a simple GA written in R to solve the simple one max problem. Here is my solution

I do much of my Bayesian computation using WinBUGS, but I also hand code samplers when I need to. Here are two examples.

-A home grown metropolis sampler MH_Norm: fort estimating the parameters for a normal distribution from a vector of random normal numbers and then an example with linear regression

-Sample code from my hierarchical time series models, this is an example that runs on the VACC again for Chironomidae

Bayesian Resources

A few blogs that I read regularly are those by Jim Albert, and Andrew Gelman.

I also have used the following books to help teach myself

Jim Clark's Models for Ecological Data

Jim Albert's Bayesian Computation in R

And a couple books by Andrew Gelman:

Data analysis using regression and multilevel/hierarchical models

Bayesian data analysis

Of course I also rely on my adviser Nick Gotelli's book A primer of Ecological Statistics

Retrieved from "https://openwetware.org/mediawiki/index.php?title=Hart:Stats&oldid=399853"