When I use "fit OneClassSVM * kernel="rbf" nu=0.5 coef0=0.5", I find that the isNormal field of some of the items display nothing. How can I deal with it?
The fit command is likely dealing with missing values during the fitting process where columns with missing values do not get predictions. Can you try doing
| table *
| fit OneClassSVM*
and confirm that the isNormal field is only empty for rows with missing values?
If that is the case, one way to get around it is by specifying which fields you want:
index=foo fieldone=* fieldtwo=* ...
or you can consider using the fillnull command:
which will fill in zeros for missing values.