Average doesn't change with deletion of tuples?

Question

Hi there, I am using Oracle 11g EE. I'd appreciate if you could shed some light on my confusion. The following statement attempts to delete all tuples in the ACCOUNT table when the balance falls below the average balance in the ACCOUNT table:
delete from account where balance < (select avg (balance) from account);
I was taught that there is an issue with this because as we delete tuples from the table, the average balance changes. I ran this statement and encountered no issues.
It correctly deleted the tuples where balance is less than the initial average of the table.
The solutions suggested to fix the potential issue are as follows:
1. First, compute avgbalance and find all tuples to delete
2. Next, delete all tuples found above (without recomputing avg or retesting the tuples)
My question is, does the average really get changed in the (select avg (balance) from account) clause as we delete tuples?
I know how to solve this potential issue. Just don't quite understand why the statement in question works perfectly. I tried testing it out by inserting tuples that have balance close to the initial average. No issues there.
Thank you in advance!

Answer 1

There's even more basic concepts involved
read consistency
A consistent view of data seen by a user. For example, in statement-level read consistency the set of data seen by a SQL statement remains constant throughout statement execution.
data consistency
A consistent view of the data by each user in a multiuser database.
SQL Processing for Application Developers
SET TRANSACTION
Regards
Etbin
something to play with
select rid,x,overall_average,running_average,
       case when case when x < overall_average
                      then 'higher'
                      when x > overall_average
                      then 'lower'
                      else 'equal'
                 end !=
                 case when x < running_average
                      then 'higher'
                      when x > running_average
                      then 'lower'
                      else 'equal'
                 end
            then case when x < overall_average
                      then 'higher'
                      when x > overall_average
                      then 'lower'
                      else 'equal'
                 end || ' ~ ' ||
                 case when x < running_average
                      then 'higher'
                      when x > running_average
                      then 'lower'
                      else 'equal'
                 end
       end average_mismatch
  from (select rid,x,
               avg(x) over (order by rid rows between unbounded preceding and unbounded following) overall_average,
               avg(x) over (order by rid) running_average
          from (select level rid,100 + trunc(10 * dbms_random.normal) x
                  from dual
                connect by level <= 20
RID
X
OVERALL_AVERAGE
RUNNING_AVERAGE
AVERAGE_MISMATCH
1
105
99.65
105
lower ~ equal
2
111
99.65
108
3
100
99.65
105.333333333333333333333333333333333333
lower ~ higher
4
102
99.65
104.5
lower ~ higher
5
108
99.65
105.2
6
87
99.65
102.166666666666666666666666666666666667
7
105
99.65
102.571428571428571428571428571428571429
8
114
99.65
104
9
85
99.65
101.888888888888888888888888888888888889
10
100
99.65
101.7
lower ~ higher
11
100
99.65
101.545454545454545454545454545454545455
lower ~ higher
12
109
99.65
102.166666666666666666666666666666666667
13
88
99.65
101.076923076923076923076923076923076923
14
89
99.65
100.214285714285714285714285714285714286
15
81
99.65
98.9333333333333333333333333333333333333
16
108
99.65
99.5
17
103
99.65
99.7058823529411764705882352941176470588
18
99
99.65
99.6666666666666666666666666666666666667
19
93
99.65
99.3157894736842105263157894736842105263
20
106
99.65
99.65
Message was edited by: Etbin
an example added