Seems like there is an issue with reltuples showing twice the numberof rows

I am relying on reltuples on my web app to get fast row counts.

However for some table I am getting twice as many values when I try to do this. I did some more research and came up with this query.

select reltuples, n_live_tup, n_dead_tup

              from pg_stat_user_tables join pg_class using (relname)

             where relname = 'main_csv_0f247511f5c247c8813ba3cec90c6ae1_yf34fbb38d';

it returns

 reltuples  | n_live_tup | n_dead_tup

-------------+------------+------------

2.7209e+06 |    1360448 |    1360448

If I run analyze main_csv_0f247511f5c247c8813ba3cec90c6ae1_yf34fbb38d

and I run the same query again,

But after some time the value goes back to being double the value. This is causing me a lot of problems since this inaccuracy does not make any sense to me.

Any help would be appreciated.

FYI, also asked the same question on stackoverflow since I am new to postgres mail lists.