OPEN SOURCE SOFTWARE MAINTENANCE EFFORT ESTIMATION: A SYSTEMATIC MAPPING STUDY

Software maintenance activities are considered as the most expensive ones within the software lifecycle. Software engineering researchers have strived to improve maintenance effort estimation (MEE) of Open Source Software (OSS) through many empirical studies for maintenance effort estimation in open source software (O-MEE). This study objective is to review the published studies in O-MEE to summarize the existing body of knowledge of this research topic. We performed a systematic map of the empirical studies on O-MEE published from 2000 up to June 2020. The 65 selected primary studies were analysed and classified according to their publications' years, channels and venues, OSS projects used as datasets, research approaches, estimation techniques, metrics used as independent variables as well as dependent variables (e.g., maintenance effort). The findings of this mapping study revealed that researchers have paid a considerable amount of attention to O-MEE in the last decade. Moreover, information on effort being rarely directly available in OSS, researchers have used indirect effort substitutes as dependent variables. The most used estimation techniques were Regression Analysis, Bayesian Networks and Decision Tree. We identified two promising emergent approaches in O-MEE: machine learning technique and estimation of the effort indirectly based on size measures such as lines of code and function points.