Читать книгу Sports Analytics in Practice with R - Ted Kwartler - Страница 17

In the end, R should be known as a scripting language. It is not a “low level language” like Java where the code itself is executed directly on the hardware such as a central processing unit, CPU. As a scripting language R cannot be compiled into an executable standalone program. As a result, R has some constraints, in that it is slower than other languages like Java and building a standalone application is not possible. However, the benefit is that R is capable of executing multiple operations borrowing from languages as needed and it well suited to statistical tasks. For example, operations in machine learning like Random Forest are called with R functions yet executed in Fortran. Still other functions borrow from C, SQL, Weka, and so on. This diversity makes the functionality high because the functional tool set is vast but with so much going on under the hood, R can be slow. Another drawback to R is that objects are stored in active memory. As a result, data objects are limited by the amount of RAM on your laptop or R-Studio server. For many data tasks this is not a problem especially as you transition to scalable cloud servers. In fact, in this text, all the data tasks and computation are relatively small. Constraints start to exist after loading millions of rows and columns in a data frame or when doing computationally complex calculations as is the case with Depp Neural Networks. However, this book is focused on sports analytics rather than big data or machine learning exclusively. Thus, these drawbacks to the R language will not be an issue.

Besides the ability to execute functions drawing from across multiple specialized languages R has other positive benefits. For example, R has a well-developed support community. Often when you are presented with an error or unknown operation, a simple online search will identify the solution. Additionally, R is optimized for statistics. Although python, a competing and more-diverse language has similar functionality, there are still differences between the two languages. For example, some machine learning-tuning parameters are better executed in R, the simple creation of dynamic HTML-based dashboards is easier or file formats like “fst” are more enhanced within R. Given that R is not picky about spacing and indentation, it is an excellent language for the novice programming. As you scale your learning in analytics and coding, you will likely want to add to your language toolkit.