r - Attribute information gain -


i r user , interested in finding out significant/important attributes/predictors classifier naive bayes multinomial package rweka.

i aware of function infogainattributeeval. function uses filter approach attribute selection. classifier independent.

i classifier dependent evaluation. tried wrappersubseteval. here did:

library("rweka")  dt <-   data.frame(     sex = as.character(c(0, 0, 0, 0, 0, 1, 1, 1, 1, 1)),     att1 = as.character(c(4, 4, 3, 4, 2, 3, 2, 3, 4, 2)),     att2 = as.character(c(4, 4, 4, 5, 3, 7, 6, 7, 8, 9)))  wse <- make_weka_filter("weka/attributeselection/wrappersubseteval")  wse(sex ~ att1 + att2,     data = dt,     control = weka_control(       b = list("weka.classifiers.bayes.naivebayes"))) 

i error:

error in .jcall(filter, "z", "setinputformat", instances) :    method setinputformat signature (lweka/core/instances;)z not found 

i have no idea error. weka problem? or r problem? or syntax error?

thank in advance help. still have kind assistant if have insight problem.


Comments

Popular posts from this blog

google api - Incomplete response from Gmail API threads.list -

Installing Android SQLite Asset Helper -

Qt Creator - Searching files with Locator including folder -