I am using teradataR (1.01) to do some clustering analysis, but I found the td.kmeans function can only receive a full table as the input.
So my question is can I assign some columns to td.kmeans and make it do analysis only on these columns?
like this way:
> test <- td.data.frame("test", "testdb")
> td.kmeans( cbind(test["x"], test["y"]) )
Can't test it right now as teradataR is not working for R3.0.0 :-(
What should work is that you create a view with your two columns and call td.kmeans with this view.
Actually the td.kmeans function takes a td.data.frame as input and td.data.frames can be used almost identically to a regular R data.frame. cbind won't work because it doesn't know about td.data.frames. You can use the following:
td.kmeans(test[1:2]) //Use first two columns
td.kmeans(test[c('col1','col2')]) //Use columns named col1 and col2