I'm working on a dataset in which a certain
feature is 'NaN' for
many patterns (many rows). 'NaN's come from divisions by zero like 0/0
or x/0. How is 'NaN' usualy interpreted by Weka? Is it considered as
missing/unknown data (i.e., 'NaN' = '?') ?
Yep, missing values are represented as Double.NaN. See also
weka.core.Instance and there the static method "missingValue()".
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
+64 (7) 838-4466 Ext. 5174