Missing is not zero.

Missing values can look harmless in a table. In a chart, the treatment can change the story. Replacing missing with zero invents an event that may never have happened; joining the points either side assumes a path you do not know.

The right choice depends on why the value is missing. A sensor outage, suppressed count and genuine zero are different states.

A useful check: could a reader tell the difference between “nothing happened” and “we don’t know what happened”?