Free copy of book by Hastie, Tibshirani, and Friedman

Hastie and Tibshirani mentioned in their course in Krems am Donau (Austria) that the last edition of their book "The Elements of Statistical Learning: Data Mining, Inference, and Prediction" will be freely available. It is true! You can fetch a free copy from Hastie's webpage (8.2 MB).

RExcel

I am not a fan of MS Windows software. However, the reality is that majority of people use it and it is often easier for me to live with this than to convert the sorounding souls. Additionally I am find ing out lately that majority of my favourite open-source programs work on all major platforms: MS Windows, Mac, and UNIX/Linux. Then there is no problem for me to work on any of these systems. MS Excel tool is similarly very widely used by mases, but not really powerfull when it comes to data analysis. However, we must admit that it is jolly usefull for handling medium sized data and that majority of people perform analyses with Excel. For those that can not live without Excel and have some working knowledge of R I suggest to take a look at the RExcel and its demo video (28 min!). It seems to be very nicely integrated - Excel keeps tracks of all the dependencies, while R takes care of the computations. Of course you can do the same stuff in R, by reruning the script for each change, but the interactivity offered by the Excel has its own merit. Regarding the quality of graphics it is obvious that Excel plots can not match with R. But if you need some interactivity (imagine you would like to study the effect of a particular parameter on distribution density), then Excel plots are good enough.

Happy "firefox"

smoothScatter in base R

smoothScatter function is now available also in base R - in the recommended package graphics that is shipped with R. Originally this function was in the genepplotter package from Bioconductor. I really like it since it can nicely plot large datasets. Bellow is an example figure (also available at R Graph Gallery). Simply start R and type ?smoothScatter to get familiar with this function.

New version of Pedigree Viewer

Brian and Sandy Kinghorn have made quite some changes to the popular Pedigree Viewer program. In particular they added the mate-selection module. I definitely have to try it out!

Sustainable selection on body weight in minks

In several species breeders want to increase the body weight to obtain more efficient animals. Often animals are fed ad libitum to give the opportunity to animals to show their growth (as well as obesity) to a maximal extent. However, Danish scientist reported that this might not be sustainable - at least for minks. They conducted the experiment with restricted and ad libitum feeding regime and found out that minks from the restricted group were more fertile and had better utilization of feed. I wonder how does this relate to pigs, rabbits, etc.

Gelmans talk at Applied Statistics conference in Ribno (Slovenia)

I had an oppurtunity to attend the talk by the Andrew Gelman at the Applied Statistics conference in Ribno (Slovenia). It was nice to meet him "in flesh" after reading his books and blog for quite some time now. He is more "funny" and relaxed than I anticipated, but his talk was very brilliant, He mainly talked about weakly informative prior - his new secret weapon ;) Some older slides about this topic are available here.

ASD talk - Growth performance of station tested rams in Slovenia

I attended the 17th ASD congress in Abano (Italy). I had a presentation about our model development for genetic evaluation of performance tested rams on test stations in Slovenia.

Growth performance of station tested rams in Slovenia

SGD Talk - Fitting pedigree based mixed models in BUGS software

I have a talk today at the 5th SGD congress about fitting animal model (mixed/hierarchical/multilevel model with pedigree information) in BUGS. The abstract was posted before, while the talk slides are available bellow.

Fitting pedigree based mixed models in BUGS software

Changing font size of R output in Lyx Sweave

Is there a simple way of reducing the size of the font of the R output in Lyx/Sweave? It should be simple, but I can't get anything to work. Placing\huge or \tiny in front of code chunk doesn't change the output.

My reply:

The trick is in what Sweave actually does. Sweave replaces the R code with the code and results, but as Sinput and Soutput environments, i.e.,

<<>>=
1 + 1
@

Becomes

\begin{Schunk}

\begin{Sinput}
> 1 + 1
\end{Sinput}

\begin{Soutput}
[1] 2
\end{Soutput}

\end{Schunk}

Therefore, I think you need to modify the Sinput and/or Soutput tex environments
defined in the Sweave.sty file. However, I have never done anything like that. I think there are some R packages on CRAN that did some of this things - I remember seeing green output etc.