|
Abstract:
|
Metaprogramming techniques are useful for translating coding statements into a new syntax. For statistical analyses in R, metaprogramming can be used to extend the capabilities of the language, facilitate iterations in design, and improve the efficiency of applications. This presentation will demonstrate the techniques of text mining, natural language processing, and dynamic generation of coding statements that are useful in translating to a new syntax, all in the context of statistical analyses and applications. Coding translation can be achieved with a low overhead because its process depends only on the length of the statement rather than the size of the data.
A number of examples of coding translations will be discussed. Dynamic generation of a modeling formula will be developed from user-generated inputs in an interactive application. When functional outputs serve as the inputs to other models, a coding translation technique improves the running time efficiency of statistical calculations. Conversion of common methods of data analysis to more computationally efficient techniques may also be used to achieve gains in performance and adaptability in iterative design.
|
