# Finding predictions using biglm without finding errors

I'm using the biglm R package for linear regression.

click/impression is the output required. But the test data does not contain click and impression. The predict function of biglm gives the error

Error in eval(expr, envir, enclos) : object 'click' not found

I assume this is because predict tries to compute the standard errors also.

Is there a method to just obtain the predictions ? I tried assigning values to se.fit and type attributes, but I get the same error.

biglm calls model.frame which "all the variables in the formula are included in the data frame" see documentation for model.frame. This is the issue that comes up when predict is called on the biglm class. It looks for those values in the predict function. To get around this you can just create a variable and encode it with 0. See below...

data(trees)
ff<-log(Volume)~log(Girth)+log(Height)

chunk1<-trees[1:10,]
chunk2<-trees[11:20,]
chunk3<-trees[21:31,]

a <- biglm(ff,chunk1)
summary(a)

#produces same error
chunk2 <- select(chunk2, -Volume)
predict(a, chunk2)

#Fixed
chunk2\$Volume <- 0
predict(a, chunk2)