hpcc dataset and R data frame mapping

[balajisampath]

hpcc dataset gets converted to data frames when passed in as parameter but, all columns in dataframes are as list. e.g. cyl becomes a list instead of vector producing the error

System error: 0: Rembed: Rcpp error: invalid type (list) for variable 'cyl'

Code: Select all

IMPORT R,ML;
mtcarsrec := RECORD
  real8     mpg;
  unsigned1 cyl;
  real8     disp;
  unsigned2 hp;
  real8     drat;
  real8     wt;
  real8     qsec;
  boolean   vs;
  boolean   am;
  unsigned1 gear;
  unsigned1 carb;
END;
hpccmtcars:= dataset([{21.0,6,160.0,110,3.90,2.620,16.46,TRUE,FALSE,4,4}],mtcarsrec);

STRING run_r(dataset(RECORDOF(hpccmtcars)) ds):=EMBED(R)
rslt=  function (ds)
{
    my_modl = lm(cyl ~ ., data = ds)
    coef(my_modl)
}
rawToChar(serialize(rslt(ds), NULL,ascii=TRUE),multiple = FALSE)
ENDEMBED;
run_r( hpccmtcars );

please help if I am missing something

[richardkchapman]

Unfortunately I'm not really very familiar with R, and in particular the differences between lists and vectors and when you would want one rather than the other. Ideally the R embed plugin would have been written by an R expert (but we don't have one).

If someone can help me understand the differences and why a vector is what is wanted in this situation, I can take a look at changing it (though I'll also have to be a little careful that I don't break any existing code).