I work on multiple statistical platforms to optimize the efficiency.
R is my main statistical software for various tasks, from basic data treatment to simulations and optimizations. As an open source free software, R provides me great versitality and flexibility in coding. With the contributions of all the R users all around the world at the R forums, it is very easy to find help in R. The packages system in R is a very powerful knowledge pool that helps you to solve almost any kind of specific problems. Most importantly, the algorithms used in the packages are open knowledge to all the users, so that we get to know exactly what we are running. R is also capable of work with multiple datasets and functions at the same workspace. Generally, I seek solutions in R first when some empirical tasks are proposed to me.
|
Stata is another powerful statistical software acclaimed by many researchers, which is capable of performing nearly all genres of statistical tasks. However, to me, it is most powerful in data management and data mining. Stata has excellent efficiency in proccessing large datasets, especially those with extremely large dimensions. For instance, the Survey of Consumer Finance has more than 3000 variables and 20000 observations, which makes it really hard to proccess in R. But in Stata, treating such large data is actually easy and quick. Stata is also great in doing simple regressions and table and graphic outputs.
|
Matlab is as powerful as R if not more, which can performe well in nearly all the statistical tasks. Moreover, Matlab is capable of solving analytical problems, which makes it more than just a statistical software. The simulation performance and high dimensional plotting is excellent in Matlab. However, Matlab is a closed ecosystem run by a profit company. Data transition from matlab to other platforms is not as pleasant as it should be. After all, I use matlab for some really specific tasks, which is known to be much more efficient in Matlab. And I have sufficient knowledge in Matlab to perform all the basic statistical tasks, but slightly more time consuming for me.
|