Exception profiling method
Each record in the input data set is processed as it would be for normal exception processing. However, instead of checking each field for exception, the value of the field is recorded. When all records have been processed, you can use the statistics in the profile report to determine the thresholds.
The thresholds are set to values that would generate the percentage of warning and problem level exceptions that you specified in the Exception Profiling panel (Figure 1).
For example, suppose the field QIESECT is specified in the Exception
Threshold data set for location DSNAPC5, with the operator set to
greater than (>), and the warning and problem thresholds set to
asterisk (*). The profiling criteria are specified as 5% for warning
thresholds and 2.5% for problem thresholds in the Exception Profiling
panel. If the data set is processed and 320 records are found for
field QIESECT
with the location DSNAPC5, exception
profiling sets the warning threshold to the 17th highest record. This
would generate 16 warning exceptions on average, which is 5% of the
records (as you requested). Similarly, the problem threshold is set
to the 9th largest record, averaging eight problem exceptions, which
is 2.5% of the records.