Browsing by Author "Ahmad Amjad"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    A Deep Learning Model for Efficient Effort Estimation in Scrum Projects
    (Library Information Services, COMSATS University Islamabad, Lahore Campus, 2022) Ahmad Amjad; FA18-RCS-028; LHR TP 8041; Dr. Farrukh Zeshan
    Software Development Effort Estimation (SDEE) is defined as the process of predicting the effort required to develop a software system. Effort estimation at the earlier stage in scrum project is specially a challenging task because the requirements are volatile in nature and it can be changeable at any phase of the whole project life cycle. A project manager needs estimation for making a decision and predicting the total budget. The success or failure of software development depends on the accurate estimation of efforts required to complete a project. In literature, the three basic ways are explored for efforts estimation in scrum projects which are purely model based method, the expert-based methods and hybrid-based methods. In the model-based method the data collected from similar previous project to make prediction about the new project, the expert-based method totally dependent on human expertise to make such kind of judgement whereas the hybrid method are the combination of both model based and expert based methods. Every method has different challenges but in the era of agile software development methodologies, traditional planning and software effort estimation methods are replaced to meet customer’s satisfaction in agile environments. However, software effort estimation remains a challenge. Although teams have achieved better accuracy in estimating story points effort required to implement user stories or issues, these estimations mostly rely on subjective assessments, leading to inaccuracy and impacting software project delivery. Some researchers are pointing good results by the adoption of deep learning to address this issue. Given the foregoing, this study proposes a deep learning model for story points estimationin t h e scrum projects. In this study proposed a model which used the combination of Bidirectional LSTM and Recurrent Highway Network two powerful deep learning techniques using the story points for efforts estimation in the scrum projects. The proposed model is trained over a large d a t a s e t s for story points estimation made by 16 open-source projects which have 23313 issue that are large story point datasets according to my knowledge. The performance of the proposed model has validated through the evaluation techniques such as Mean absolute error (MAE), Mean square error (MSE), and Median absolute error (MdAE). The results shown that the proposed model provides betters result as compared to previous approaches