ORE: invalid cost.matrix column names

Question

I am trying to use the odmDT method with applying custom costs like so:
> dt.mod = ore.odmDT(default ~ .,credit_train,max.depth=20,cost.matrix=matrix(c(0,1,4,0),nrow=2)
but I get the error:
Error in ore.odmDT(default ~ ., credit_train, max.depth = 20, cost.matrix = error_cost) :
  ORE: invalid cost.matrix column names
I can't find any documentation anywhere on how exactly this cost matrix needs to be specified.
Any help would be appreciated

Answer 1

The value for cost.matrix currently requires a data frame object. The following examples using the NARROW data successfully pass a data frame for the value of cost.matrix:
    R> NARROW.costs.gender <- data.frame(ACTUAL_TARGET_VALUE = c("M","F","M","F"),
                                      PREDICTED_TARGET_VALUE = c("M","M","F","F"),
                                      COST = c(0.0, 0.25, 0.75, 0.0))
    R> NARROW.costs.class  <- data.frame(ACTUAL_TARGET_VALUE = c(0,1,0,1),
                                      PREDICTED_TARGET_VALUE = c(0,0,1,1),
                                      COST = c(0.0, 0.25, 0.75, 0.0))
    R> dt.mod1  <- ore.odmDT(GENDER ~ ., NARROW, cost.matrix=NARROW.costs.gender)
    summary(dt.mod1)
    R> dt.mod2  <- ore.odmDT(CLASS ~ ., NARROW, cost.matrix=NARROW.costs.class)
    summary(dt.mod2)
I've filed a bug so this behavior may be evaluated for a future Oracle R Enterprise release.