libRDataFrame
A library, containing the consolidated collection of Data Scientific routines to support R-style Data Structures for C++
Overview and Statment of purpose (goal development)
see Project page
This project has the goal of developing an R-like dataframe class and other data structures (like an R factor data type (called a category for C++) for similar applications for analysis of data written entirely in C++.
The necessary building blocks of this dataframe analog are a Variable composite data structure including vector class for the raw data, for the columns of the dataframe, and a vector class to contain the variable objects. Of note, R arrays are column major memory structures, so this is simulates such a structure to provide at least a virtual variable oriented structure.
The R dataframe is, a list derivative with vectors for the columns as elements of the list.As noted below, this is a bit untenable in C++. The base reference to the elements (rows and columns) is the usual numeric endices. However, associative reference with variable (column) and observation (row) also a feature. This is accomplished with a vector< vector< Node> > > structure.
Being relatively new to C++, this is my first major project in C++. At this point, my breadth and depth of understanding of C++ is growing but is at times an impediment to finding the best way to do things. Each revision provides new experience. Suggestions and comments are welcome.
Contributors
- Author: David York (Medmatix)
Development Notes
A Todo List
- Decide on the on going approachs to a mixed type list.
- A complete set of getters and setter as will as appropriate to_string methods for each datatype constructed. Include formatted display methods for each structure.
- Node class completed getters and setters
- Category class on going development, see individual project
- UtilCSV completed getters and setters
- RDataframe class getters and setters in process, basic ones are complete
- Develop a list of functions member and non-member, to include as dataframe manipulator and analysis functions.
- Continued testing ( main() module) of functions and classes as developed.
Design directions
Initially a list of vectors was the basic plan in keeping with the R structure. However R lists and C++ STL lists are different enough as to handicape any attempt to mimic R structures in C++. Thus a move tp the data.frame in C++ being a vector of vectors.
The principal issue is the need to random access the variables (columns) in the data.frame is the C++ implementation. The C++ list is implemented as a sequentially accessed object to allow access for actions on the elements across the list. The random access of individual list members is not implemented in the list structure.
Unit Testing Notes
Excluding the category data type class, all the relevant getters, setters and display functions are implemented and tested in the Unit Test main() module.
References and Acknowledgements
See related project, UtilCSV RDataframe7
Contacts
Author web page, http://crunches-data.appspot.com Email, medmatix@gmail.com
Copyright 2016 D.A.York