Strong validation on R Wrapper

[manuparra]

Validate all fault points:

In this example:

ro.r("""
        dataset = read.csv(file="{0}", header = TRUE)
        resultfit = lm(formula="{1}", data=dataset,{2})
        save(resultfit, file="{3}")
        ... Add PMML GENERATE...
     """.format(dataset, formula, inputparameters, output))

Possible fault points are:

• Reading the DataSet (remember dc:// path ):
  • What if dataset is not available
  • What if dataset has not HEADER,
  • What if formula doesn't contain columns in the dataset,
  • more ...
• LM function lm(.... :
  • Formula is not correct
  • Formula contains fields not in dataset
  • Function lm return ERROR ? or OKAY (a model)?
  • more ...
• Save result and PMML generate:
  • resultfit is OKAY (a model) or is an error?
  • file is okay and can be stored ?
  • more ...

[Mancuerna]

Done.