Stat Mach Consulting - Pawel Kobylinski, Ph.D.

21/10/2014

R and experimental design: I am sharing *wide2long* function, which I wrote for friendly conversion of repeated measures data - from wide into long format required by R packages

The function introduced here, written and named “wide2long” by myself, is basically a wrapper over the “melt” function from the “reshape2” package by Hadley Wickham. It adds some functionality, though. The wrapper is meant to be just a little bit more friendly to experimental scientists than the original “melt” function. It is also meant as a basic building block - a sub-function - of functions which I will describe in next posts. “wide2long” handles a very simple design (a special case of A/B testing), in which there is only one repeated measures variable. Functions coming in next posts will deal with more complex situations.

08/10/2014

LASSO regression with julia and R - quick elegant code chunks can be found here. LASSO is one of the methods that help in variable count reduction, when dealing with big data.



http://statcompute.wordpress.com/2014/10/07/fitting-lasso-with-julia/

Julia Code R Code

06/10/2014

Data Science & Digital Culture - welcome to K 01 10, a blog just launched by myself and Riam Kanso as a result of Warsaw-London intellectual chemistry



http://www.k0110.com/2014/10/welcome-to-k-01-10.html

17/09/2014

Again on the need for interdisciplinarity in Big Data. This time the American Statistical Association calls for it.

The American Statistical Association says experts in the most mature data science are underutilized by most organizations conducting data research.

02/09/2014

Why Data Science Needs Social Science. And Vice Versa
In this post for PERSONTYLE I am explaining both the kinship data science has with social science and the need for mutual transfer of know-how between the fields.

Significance of social science to perform meaningful data science and vice versa. When disciplines crash into each other great stuff can emerge.

10/08/2014

Vectorization in R (or how to speed up your R code) - short introduction with links to more detailed resources - easy read

Here are my notes from a recent talk I gave on vectorization at a Davis R Users’ Group meeting. Thanks to Vince Buffalo, John Myles White, and Hadley Wickham for their input as I was preparing this. Feedback welcome!

15/06/2014

Cluster analysis in R: determine the optimal number of clusters - consise, utterly comprehensive list of methods, with emphasis on vizualization and resources (possibly best ever answer on Stack Overflow ;) )

Being a newbie in R, I'm not very sure how to choose the best number of clusters to do a k-means analysis. After plotting a subset of below data, how many clusters will be appropriate? How can I pe...

06/06/2014

On the Deviance Information Criterion (DIC) - a measure of both fit and complexity of a model. With citation of the original paper.

09/05/2014

"No adjustments are needed for multiple comparisons" by Rothman. This classic (1990), brief paper shows how elusive and/or controversial is one of the most wide-spread, basic "axiom" of statistical and research methodology. Great read :)



http://psg-mac43.ucsf.edu/ticr/syllabus/courses/9/2003/02/27/Lecture/readings/Rothman.pdf

01/05/2014

A list of both multivariate and univariate *information measures* with resources. Very informative post, concise and up-to-date.

Malka Gorfine writes: We noticed that the important topic of association measures and tests came up again in your blog, and we have few comments in this regard. It is useful to distinguish between the univariate and multivariate methods. A consistent multivariate method can recognise dependence betw…

27/04/2014

MapReduce with R on Hadoop and Amazon EMR

and well... ok...

You all know why MapReduce is fancy – so let’s just jump right in. I like researching data and I like to see results fast – does that mean I enjoy the process of setting up a Hadoop cluster? No, … Continue reading →

27/04/2014

For teachers by teachers: statsTeachR - an open-access, online repository of modular lesson plans in R

statsTeachR is an open-access, online repository of modular lesson plans, a.k.a. "modules", for teaching statistics using R at the undergraduate and graduate level. Each module focuses on teaching a specific statistical concept. The modules range from introductory lessons in statistics and statistic…