Mar 5, 2015

Reproducible research

Since the recommendation of my professor, Riccardo Rigon, I have been thinking about reproducible research. He always insist on approaches to reproduce our activities (researches), e.g here.  I have been studying how to do it and recently I have been following Reproducible research webinar course  on coursera. What I have discover, I was not aware of this,  is that RStudio has enough resources and packages to reproduce our data analysis.

The practicality of reproducible research is really very wide and needs good skills  in database, cloud storage and version controlling tools, IDE, markup languages, and/or integration of those. However, it seems for me that it is relatively easy to start with the data analysis reproducibility i.e to take care how the data analysis is done in RStudio.  The RStudio has knitr package for this job and markup language for formatting.

I always have problem on how to handle the codes (both I steal and I did)  and forget how I did my analysis (both modelling simulations and the data analysis on R). This reproducible could save me from this trouble, and can also develop a sense of controlling what is done. It seems that it could also increase efficient because the approach we like is already there with enough documentation. So I decided to start to do my research (at least my data analysis in R) in reproducible research principles. Finger crossing!