Sorted by DateClassified by Publication TypeSorted by First Author Last NameClassified by Topic

Better, Faster and Cheaper: What is Better Software?

Burak Turhan, Çetin Meriçli, and Tekin Meriçli. Better, Faster and Cheaper: What is Better Software?. In PROMISE 2010: The 6th International Conference on Predictive Models in Software Engineering, 2010.

Download

[PDF] 

Abstract

Background: Defects are related to failures and they do not have much power for indicating a higher quality or a better system above the baseline that the end-users expect. Nevertheless, defect counts are commonly used as measures to capture the quality of a system. Further, the statistical association between internal design metrics and quality in terms of defects has been shown in previous work. Aims: Our goal is to conduct an initial data analysis for our longer-term goal of investigating whether there exist a similar relationship between the internal design characteristics of a system and quality perception of end-users. Method: We carry out an exploratory case study in robotic soccer domain. We propose a quality measure derived from the performances in the robotic soccer competitions. Then, we compare the design characteristics and quality levels of two cases. In particular, we compare the two different implementations of a system in terms of their design metrics and their overall quality as measured by game scores. Results: There are statistically significant differences between the two implementations in six out of seven design metrics. The implementation that has a much larger code base shows significantly better design characteristics in terms of complexity. We see significant differences in the quality levels as well. Conclusions: We observe that the implementation that has achieved a better performance in the tournament also has better design characteristics for almost all attributes. However, our analysis does not include enough data points to investigate any association between internal design metrics and the proposed quality metric. Another restriction is that the proposed metric is domain specific. In future work, we plan to extend our analysis using additional implementations of the same system to test the consistency of the results.

BibTeX

@inproceedings{turhan2010a,
  author    = {Burak Turhan and Çetin Meriçli and Tekin Meriçli},
  title     = {Better, Faster and Cheaper: What is Better Software?},
  booktitle = {PROMISE 2010: The 6th International Conference on Predictive Models in Software Engineering},
  year      = {2010},
  abstract  = {Background: Defects are related to failures and they do not have much power for indicating a higher quality or a better system above the baseline that the end-users expect. Nevertheless, defect counts are commonly used as measures to capture the quality of a system. Further, the statistical association between internal design metrics and quality in terms of defects has been shown in previous work. Aims: Our goal is to conduct an initial data analysis for our longer-term goal of investigating whether there exist a similar relationship between the internal design characteristics of a system and quality perception of end-users. Method: We carry out an exploratory case study in robotic soccer domain. We propose a quality measure derived from the performances in the robotic soccer competitions. Then, we compare the design characteristics and quality levels of two cases. In particular, we compare the two different implementations of a system in terms of their design metrics and their overall quality 
as measured by game scores. Results: There are statistically significant differences between the two implementations in six out of seven design metrics. The implementation that has a much larger code base shows significantly better design characteristics in terms of complexity. We see significant differences in the quality levels as well. Conclusions: We observe that the implementation that has achieved a better performance in the tournament also has better design characteristics for almost all attributes. However, our analysis does not include enough data points to investigate any association between internal design metrics and the proposed quality metric. Another restriction is that the proposed metric is domain specific. In future work, we plan to extend our analysis using additional implementations of the same system to test the consistency of the results.},
  bib2html_pubtype = {Refereed Conference},
  bib2html_rescat = {Software Engineering},
  bib2html_dl_pdf = {../files/turhanPROMISE2010RoboCup.pdf},
}

Generated by bib2html.pl (written by Patrick Riley ) on Thu May 01, 2014 16:27:43