An interface for node.js to statistical programming language R based on the fabulous Rcpp package
An interface for node.js to statistical programming language R based on the fabulous Rcpp package
Currently, rstats
is ONLY supported for Unix operating systems.
Also, it is required that the R packages RInside
, Rcpp
and RJSONIO
are installed inside R. Additionally, building the package using node-gyp
requires
python
(v2.7
, v3.x.x
is not supported)make
With these prerequisites satisfied, one can simply install rstats
using npm
npm install rstats
After installation, the package can be loaded as follows:
var rstats = require('rstats');
Once the package is loaded, we can create an R session by the command
var R = new rstats.session();
Evaluating R expressions is easy. We can use the parseEvalQ function as follows:
R.parseEvalQ("cat('\n Hello World \n')");
Numeric values can be easily assigned to variables in the current R session:
R.assign('x', 17);
R.assign('y', 3);
// calculate the sum of x+y and print the result
R.parseEvalQ("res = x + y; print(res);");
To retrieve an object from the R session, we use the get command. For example, let us create a 2x2 matrix in R and retrieve it in JavaScript as a nested array:
R.parseEvalQ("mat = matrix(1:4,ncol=2,nrow=2)");
var mat = R.get('mat');
Internally, the get function uses JSON in order to convert the R data types to JavaScript data types.
We can also run much more complicated calculations and expose the R objects to JavaScript. Consider a linear regression example:
R.parseEvalQ('x = rnorm(100); y = 4x + rnorm(100); lm_fit = lm(y~x);');
var lm_fit = R.get('lm_fit');
var coefs = lm_fit.coefficients;
var residuals = lm_fit.residuals;