Regression
WELIS offers a state of the art regression testing tool to ensure that insurance
and annuity values produced by the calculation engines do not change accidentally
as maintenance is performed. The regression tool can also compare presentation output
to benchmark values in order to maintain the integrity of text and presentation
formats.
Regression
Regression can test numeric output against benchmarked values, text output, or both.
If values have changed, a red "X" will appear in the appropriate column
on the left (either in the "Numbers" column, the "Text" column,
or both).
If a red "X" appears, changes can be seen by clicking the "View Output"
button on the Regression toolbar. Output pages will appear on a split screen, with
the benchmark values on the left and the new values on the right. Differences highlighted
in yellow.
To view the particulars of a case stored in Regression, click the "View Input"
button on the toolbar. Regression will call the Ascent User Interface, and the case
will appear loaded into the illustration system. The case can then be run or modified
as usual. Case files can also be imported and exported as standard XML files. Benchmarked
values run by other illustration systems can be imported with our regression tool
as well, saving the time and cost of duplicate case analysis.
Individual cases that share certain input or output similarities can be identified
by applying a filter to the database. Filters can be single criteria or any combination
of input or output criteria. When the filter is applied to a list of cases in a
database, only those cases that meet the filter's criteria will appear to the user.
These cases can then be selected and run, saving time and effort in the process.
Regression tracks the name of the last user to change the inputs of a case. Users
can also apply a descriptive tag to each case. Benchmarked values can include annual
or monthly values, fiscal quotes, separate fund "bucket" values, or non-ledger
values such as SERP or Deferral. Regression databases can hold hundreds or thousands
of stored cases - there are no limits to this testing tool.