I would like to start reporting defect density, however we are testing an Oracle ERP system where we about 50/50 core functionality to custom code. While we could probably get the lines of code numbers for our custom stuff, there's no way possible to get that for the core. So, that makes using the traditional defect density calculation of # defects over lines of code a bad measure since we cannot report it for 50% of our application.
The only consistency that we have is our tests. How many tests we have identified/created by feature/function. So I was looking at doing a defect density using the number of defects over the number of tests but was wondering if there was a different name for this measure than Defect Density. Personally, I would just call it Defect Density and define it how I want and be done with it, but there is a push towards standards going on here and I know someone (probably many) would google Defect Density and then try to call me out on it.
Again, Iím just trying to see if there is already a predefined measure out there to do what I am trying to do here.
Always remember that you are unique. Just like everybody else.
Let me ask - what's your goal. Do you want to measure anything just to have something measured; do you have a guideline that say bugs must be measured?
So about your measure ... could you clarify what measure exactly are you going to collect? Am I right that you will divide software into list of items (features and function) and for each item count number of test cases and number of defects. When I collected metrics like that I called them "test case effectiveness measurement metrics" reflecting the goal. I did it for regression tests only and included number of times we executed each test case per release back then.
More questions (issues I recall from back then): does each test case and each defect map to exactly one item? Can't it be that a defect cause is integration between them? Do you have test cases that test integration between them? Will you include defect weight (severity) and test case weight (complexity, time to execute) or just count them?
Note that the measure is completely different from defect density. Defect density aims to measure coding while mine aimed to measure testing itself.
?:the art of a constructive conflict perceived as a destructive diagnose.