Game Development Reference
Another useful software QA method is to plot control charts of product or process
results. The control chart shows the average result to expect and “control limit�?
boundary lines for the set of data provided. Any items outside of the control limits fall
beyond the range of values that would indicate they came from the same process as
the rest of the data. This is like having a machine that stamps metal squares a certain
way, but every once in a while, one comes out very different from the others. If you
have the right amount of curiosity to be a QA person, you would want to know why
the square comes out wrong some of the time. The same is true for software results
that come out “funny.�? The control chart reveals results that should be investigated to
understand their cause. It might simply be a result of someone entering the wrong
data (date, time, size, defects, and so on). Figure 6.8 shows an example control chart
for new lines of delta (added or deleted) code changes in the game each week. The
numbers are in KLOC.
Figure 6.8 Control chart of weekly code change in KLOC.
The solid line running across the middle of the chart is the average value for the data
set. The two dashed lines labeled UCL and LCL represent the Upper Control Limit
and the Lower Control Limit, respectively. These values are calculated from the data
set as well. The data point for the week of 5/2/2004 lies above the UCL. This is a point
that should be investigated.