Effort estimation is a crucial part of software project planning.  Improper estimates can lead to consequences ranging from delays to project cancellation.  In a survey of the field, it was found that most projects (60-80\%) encountered effort overruns, schedule overruns, or both. \cite{1237981} 
\\
In order to address the need for better software project planning methods, the field of search based software engineering has been developed.\cite{Harman2001833}  The field suggests stochastic search processes can be applied to existing algorithms in order to improve performance of algorithmic estimation techniques.  Search based software engineering has grown since its inception, but there are still many open research questions.  Harman suggests that a promising area of future work is finding a stopping criteria that relies on comparison between the similarity of proposed solutions.\cite{Harman2007}
\\  
This requires an effective way of comparing the solutions two algorithms generate, to assess their similarity.  There are numerous ways to approach this, as there is no standard performance measure for evaluating effort estimator performance.  This has been a criticism against the field of empirical software engineering\cite{eS} such that better performance measures could also have an impact in the larger software engineering community.
\\
This paper seeks to address both concerns, by analyzing different performance measures impact on algorithm rankings.  This is done using a combination of algorithms approach (COMBA), in which a data preprocessor and learner and combined.  This creates a large number of test algorithms from a smaller number of preprocessors and learners, creating a wide range of example solutions for testing ranking methods.  This approach has been applied to effort estimation previously, in which it was found that certain datasets distinguish between algorithm performance more strongly and thus more useful for optimizing effort estimator performance.\cite{ICSE}  This paper will additionally try to reproduce and verify the results of the previous study.
\\
\noindent{The paper is laid out as follows:}\\
Section 2 will cover previous works regarding this subject.\\
Section 3 will describe the experimental procedure used.\\
Section 4 provides the results of the experiment.\\
Section 5 concludes and suggests future works using a combination of algorithms approach.\\