setPemaClass: PEMA classes

Artikel
03/23/2017

Applies to version 8.0.3 of package RevoPemaR.

Description

Returns a generator function for creating analysis reference class objects to use in pemaCompute.

Usage

  setPemaClass(Class, fields = character(), contains = character(), 
      methods = list(), where = topenv(parent.frame()), includeMethods = TRUE, ...)

Arguments

Class

character string name for the class.

fields

either a character vector of field names or a named list of the fields.

contains

optional vector of super reference classes for this class. The fields and class-based methods will be inherited.

methods

a named list of function definitions that can be invoked on objects from this class.

where

the environment in which to store the class definition.

includeMethods

logical. If TRUE, methods (including those of parent classes) will be included when serializing.

...

other arguments to be passed to setRefClass.

Details

See setRefClass for more information on arguments and using reference classes. The setPemaCLass generator provides a framework for writing parallel, external memory algorithms (PEMAs) that can be run serially on a single computer, and will be automatically parallelized when run on cluster supported by RevoScaleR.

Value

returns a generator function suitable for creating objects from the class, invisibly.

Examples



 # A very simple example of computing a sum
 PemaSum <- setPemaClass("PemaSum", 
   contains = "PemaBaseClass",

   fields = list( 
       # list the fields (member variables)
       sum = "numeric",
       varName = "character"
       ),

   methods = list(
       # list the methods (member functions)
       initialize = function(varName = "", ...) 
       {
             'sum is initialized to 0'          
             # callSuper calls the method of the parent class
             callSuper(...)            
           usingMethods(.pemaMethods)  # Will include methods if includeMethods is TRUE        
             # Fields are modified in a method by using the non-local assignment operator
             varName <<- varName
           sum <<- 0
       },
       processData = function(dataList) 
       {
           'Updates the sum  from the current chunk of data.'
             sum <<- sum + sum(as.numeric(dataList[[varName]]), na.rm = TRUE)
           invisible(NULL)
       },
       updateResults = function(pemaSumObj)
       {
             'Updates the sum from another PemaSum object.'
             sum <<- sum + pemaMeanObj$sum
             invisible(NULL)
       },
       processResults = function()
         {
             'Returns the sum. No further computations required.'
           sum
       },
         getVarsToUse = function()
         {
             'Returns the varName.' 
             varName
         }
   )
 )

 pemaSumObj <- PemaSum()
 pemaCompute(pemaSumObj, data = iris, varName = "sepal.length")

Share via