How does druid handle columns (to be ingested as a metric) with null values?

For my parquet dataset, I have two options for certain columns(to be ingested as metrics).

  1. Add the values as null for a record.

  2. Add value “0” for a record.

While aggregating, I do not want these values to mess up the aggregated results, so which one should I choose ?

Is one a better option than the other?

Aggregating involves computing columns as numbers which means the latter is what you wanted.

Rommel Garcia