[R] predict(): NoSuchMethodError

[nstender49]

I am currently working on a code which clusters attributes from a data set,
then uses a linear regression model to predict NA values in the data set. 
The code works for almost all cases, but then errors out on a case that
seems like it should work the same. 

This is the line of code that is giving me the problem:

NewClusterData[j,att] <- predict(cl,newdata = ClusterData[j,],type="class")

Here is the code for the arguments:

algName <- "LinearRegression"
Formula <- as.formula(paste(as.name(att), " ~ ",  paste(otherAtts, collapse=
"+")))
myAlg <- get(algName)
cl <- myAlg(Formula, data = Data)

And this is the error I'm getting:

Error in .jnew("weka/core/Attribute", attname[i]) :
java.lang.NoSuchMethodError: <init>

Here is the printout of the a working case:

cl = Linear Regression Model
      petallength =
           1.748  * sepallength +
          -1.4174 * sepalwidth +
          -2.1294
   
ClusterData[j,] = sepallength sepalwidth petallength petalwidth class
                        5.7           2.6           NA            1            
1

petallength = 4.148813

And the non-working case:

cl = Linear Regression Model
      petalwidth =
           0.8948 * class +
           0.3031
ClusterData[j,] = sepallength sepalwidth petallength petalwidth class
                        4.9            3             1.4            NA          
0

Error in .jnew("weka/core/Attribute", attname[i]) : 
  java.lang.NoSuchMethodError: <init>

Thanks for any help that anyone can provide!
-- 
View this message in context: http://old.nabble.com/predict%28%29%3A-NoSuchMethodError-tp26521075p26521075.html
Sent from the R help mailing list archive at Nabble.com.