A Metaprogramming Algorithm to Simplify the Coding Syntax of Data Analysis Methods

This study details an application of metaprogramming techniques used to create an alternative coding syntax for data analysis methods in the R language. Other common techniques for data analysis include chained commands or queries with a specific structure. The author designed a software package (DTwrappers for R) that implements data analyses using functional programming. Coding statements based on functions can be easier to learn and work more generally in many contexts such as dynamic applications. The DTwrappers package includes methods that cover many common techniques for data processing and analysis, such as counting rows, assigning values to variables, selecting rows and columns of data, and performing simultaneous calculations. Many of these methods can be extended to allow for filtering of the data, grouped calculations, and specialized operations like selecting the first k rows of data in each group. For each of these methods, DTwrappers utilizes metaprogramming techniques to dynamically generate coding statements based on the data.table package for R. These translations are constructed from string processing techniques that map the parameters of the original functional statement into the appropriate form of the corresponding data.table calculation. As a result, data analysts can utilize a simplified functional programming syntax while benefitting from the optimized speed and memory efficiency of data.table. This study provides details on the methods of coding translations used to map from coding statements in DTwrappers to translated coding statements for data.table.