0

can anyone tell me what am I doing wrong in WEKA tool? I am trying to supply a testing set with the same attributes as the training set but when looking at instances or sum of weighs there is a "?". What I could possibly be doing wrong?

enter image description here

Louise
  • 83
  • 5

1 Answers1

1

If it is an ARFF dataset, then Weka will load it incrementally for testing rather than fully into memory (which can be a problem for very large test sets). Since Weka doesn't know the full extent of the dataset, it can neither display the number of instances nor sum of instance weights.

fracpete
  • 2,448
  • 2
  • 12
  • 17
  • Thank you for responding @fracpete - Do you reckon that then would be better to merge both training and testing sets in one file (e.g., csv) and do percentage split on weka tool? – Louise Aug 19 '22 at 12:50
  • 1
    That dialog is just showing you some information, which can be limited if the data type can be loaded incrementally (like CSV). It is not an error. It is usually recommended to convert datasets to ARFF, as that format specifies expkicitly its data types in yhe header. CSV files can easily be incompatible if they differ in nominal values for attributes. It is really up to you whether to have a predetermined dataset split or perform a train/test split on the fly. – fracpete Aug 19 '22 at 21:39