In general, HELP STATS doesn’t do anything to the confidence levels.
However, one possibility could be random AMP sampling. The HELP STATS operation logic triggers all-AMP random AMP sampling, which may be impacting some confidence levels.
Before the statistics dictionary table (dbc.statstbl) is checked for the existence of statistics, we first fetch the table header. The random AMP samples get brought into the dictionary cache as part of accessing the table header. Your first explain had no random AMP samples to look at, your second explain did. I can't say for sure, but that could be what happened in your case.
In general, HELP STATS doesn’t do anything to the confidence levels.
However, one possibility could be random AMP sampling. The HELP STATS operation logic triggers all-AMP random AMP sampling, which may be impacting some confidence levels.
Before the statistics dictionary table (dbc.statstbl) is checked for the existence of statistics, we first fetch the table header. The random AMP samples get brought into the dictionary cache as part of accessing the table header. Your first explain had no random AMP samples to look at, your second explain did. I can't say for sure, but that could be what happened in your case.
Thanks, -Carrie