| Summary: | 碩士 === 國立臺灣海洋大學 === 資訊工程學系 === 95 === The purpose of this thesis is to compare the accuracy of estimation model that is created by using homogeneous software system maintenance historical data and by using heterogeneous software system maintenance historical data.
Maintenance phase accounts for 40% to 80% of the software life cycle cost, planning software maintenance work is a key point for a successful maintenance project. To this aim, effort estimation is a valuable asset to maintenance managers in planning and performing feasibility and risk analysis.
By intuition, it is considered that creating estimation models with homogeneous dataset should result in higher estimation accuracy, and creating estimation models with heterogeneous dataset should result in poorer estimation accuracy. In fact, there are some researches performed estimation models with highly accuracy by using heterogeneous data sets, and some performed by using homogeneous datasets. No matter what kind of datasets ware used, all of these researches performed good result. It violates the intuition mentioned above. For this reason, we purposed a question, that is, does homogeneity and heterogeneity of dataset influence the accuracy of estimation?
Therefore, the main purpose of this thesis is to compare the accuracy of estimation models that is created by using homogeneous data and by using heterogeneous data.
|
|---|
