Software Development Effort Estimation Using Feature Selection Techniques

Context: Software Development Effort Estimation (SDEE) remains as the most important activity in software project management. Providing accurate estimates at early stages of software life cycle was the subject of a large number of studies for more than four decades. Therefore, different SDEE techniques have been proposed and evaluated. In order to improve the accuracy of the proposed estimation techniques, many researchers investigated different data preprocessing tasks in combination with SDEE techniques, especially Feature Selection (FS). Objective: Performing a systematic mapping study (SMS) of papers investigating the use of feature selection techniques in SDEE. We analyze and synthesize the selected papers according to 7 aspects: publication venues, year of publication, research type, empirical type, type of feature selection, feature selection techniques and estimation techniques. Method: A SMS was performed on the studies published in four digital libraries between 2000 and 2017. Conclusion: 45 papers were selected to answer the mapping questions. Moreover, 18 different FS techniques belonging to different categories (e.g. Filters, Wrappers and Hybrid) were investigated. The impact of the FS techniques was assessed using 9 SDEE techniques.