No abstract

Software maintenance is the process of fixing, modifying, and improving software deliverables after they are delivered to the client. Clients can benefit from offshore software maintenance outsourcing (OSMO) in different ways, including time savings, cost savings, and improving the software quality and value. One of the hardest challenges for the OSMO vendor is to choose a suitable project among several clients' projects. The goal of the current study is to recommend a machine learning-based decision support system that OSMO vendors can utilize to forecast or assess the project of OSMO clients. The projects belong to OSMO vendors, having offices in developing countries while providing services to developed countries. In the current study, Extreme Learning Machine's (ELM's) variant called Deep Extreme Learning Machines (DELMs) is used. A novel dataset consisting of 195 projects data is proposed to train the model and to evaluate the overall efficiency of the proposed model. The proposed DELM's based model evaluations achieved 90.017% training accuracy having a value with 1.412 × 10 -3 Root Mean Square Error (RMSE) and 85.772% testing accuracy with 1.569 × 10 −3 RMSE with five DELMs hidden layers. The results express that the suggested model has gained a notable recognition rate in comparison to any previous studies. The current study also concludes DELMs as the most applicable and useful technique for OSMO client's project assessment.

The successful execution and management of Offshore Software Maintenance Outsourcing (OSMO) can be very beneficial for OSMO vendors and the OSMO client. Although a lot of research on software outsourcing is going on, most of the existing literature on offshore outsourcing deals with the outsourcing of software development only. Several frameworks have been developed focusing on guiding software system managers concerning offshore software outsourcing. However, none of these studies delivered comprehensive guidelines for managing the whole process of OSMO. There is a considerable lack of research working on managing OSMO from a vendor's perspective. Therefore, to find the best practices for managing an OSMO process, it is necessary to further investigate such complex and multifaceted phenomena from the vendor's perspective. This study validated the preliminary OSMO process model via a case study research approach. The results showed that the OSMO process model is applicable in an industrial setting with few changes. The industrial data collected during the case study enabled this paper to extend the preliminary OSMO process model. The refined version of the OSMO process model has four major phases including (i) Project Assessment, (ii) SLA (iii) Execution, and (iv) Risk.

No abstract