Results
Here, with respect to research questions RQ2 and RQ3 we summarize the results of our simulation. Research question RQ1 was answered in section \ref{model_construction}.
RQ2: Will usage of the DePress framework in industrial project pay off?
Simulation shows, that we should expect Benefit (\ref{benefit}) from DePress usage in the project:
\begin{equation}
\label{benefitprim}Benefit=1985x-1393x=592x\\
\end{equation}
Accordingly, expected Return on Investment (\ref{roi}):
\begin{equation}
\label{roiprim}ROI=\frac{592x-8x}{8x}=73\\
\end{equation}
As ROI is positive, we can state that investment will pay off.
RQ3: How cost effective is the DePress framework for defect prediction in an industrial software development project with the highest possible prediction recall achievable for that tool?
As shown in Table \ref{tab:costs-table}, the expected total investment cost of DePress tool-based defect prediction application in software development project is:
\begin{equation}
\label{investment}Investment=8x\\
\end{equation}
Benefit Cost Ratio (\ref{bcr}) calculated using (\ref{benefit}) and (\ref{investment}) values:
\begin{equation}
\label{bcrprim}BCR=\frac{592x}{8x}=74\\
\end{equation}
Such BCR value shows that we should expect a high monetary gain from DePress tool usage for supporting quality assurance with defect prediction. Moreover, NetReturn (\ref{netreturn}) from the simulated defect prediction application is:
\begin{equation}
\label{netreturnprim}NetReturn=592x-8x=584x\\
\end{equation}
Answering research question RQ3, when the defect prediction application strategy proposed in section \ref{strategy_par} is applied, such an approach can result in reduction of final QA costs by almost 40%, when recall of the prediction model will be 0,952, and after fixing 76% of the detectable bugs (\ref{h1prim}) while still in the first, developmental phase of the project. Graphical comparison of quality assurance costs – actual and simulated, with defect prediction introduced, is shown in Figure \ref{fig:cost_comparison}.