Abstract
[27] replicated previous studies conducted by Shepperd et al. [31], but found analogy was not better than regression, and they also suggested that the results are sensitive to experimental design. Similarly, Briand and Eman [6] and Morasca [25] found analogy-based systems were less robust than other methods, particularly when dealing with heterogeneous data sets. Jeffery and Ruhe [12] also concluded stepwise regression outperformed analogy-based systems with the ISBSG dataset. Over the past 10 years, researchers tried to explain why different research teams have reported widely different results by using analogy. Most recently, Mair and Shepperd [22] undertook a systematic review to investigate these contradictory results. They reviewed 20 primary studies comparing regression and analogy conducted during the past decade, and concluded that there was no clear indication that regression was better than analogy or vice versa. They concluded that the mixed results are due to the characteristic of the dataset and the individual data points [29]. The implication is that the resultant prediction is sensitive to the data quality of individual datasets. Shepperd and Kadoda [29] have also studied this issue using simulation and arrived the same conclusion. As suggested by Mair and Shepperd [22], we agree researchers and practitioners should be focusing on when to use technique A rather B, as opposed to attempting to identify one single method as the best method. What is important to a project manager is which method is most appropriate in his/her specific circumstances as many of these methods are context-specific, and what information and data are immediately available. Despite of its great significance in the software cost estimation research, estimating by analogy however remains largely unfamiliar by many industry software developers today, especially developers from small to medium organizations. This paper is intended for project managers and researchers wishing to gain in-depth knowledges of analogy-based methods. It provides a comprehensive overview for whom interested in learning analogy-based approaches, and a review of the underlying Software project managers req