Hi nagarjuna559,
Most fields will have a comparable effect on performance. One notable exception is categorical fields that have many distinct values, because MLTK will use one-hot encoding to convert those fields into numeric fields. For example, a field that contains US states may have ~50 different values, and that will explode into ~50 fields before the model is fit. You can try running "| stats dc" to see how many distinct values your categorical fields have (ignore the numeric fields).
If that doesn't work, please provide more information about the number of fields you have, what algorithm you're using, etc.
... View more