A Critical Review for Software Maintenance Cost Estimation Models: A Data Driven Approach

Software maintenance is a crucial phase in the lifecycle of software development model; it requires a significant financial outlay as well as resources. To effectively plan and manage a project, accurate cost estimation is necessary. Fixing bugs, optimizing performance, adding new features, and adhering to regulations are just a few of the tasks that go under software maintenance. Because software systems are dynamic and requirements change over time, it can be difficult to estimate the cost of these tasks. To help businesses anticipate and control these expenses, a number of software maintenance cost estimating models have been developed over time. Knowledge-based modeling, algorithmic modeling, and non-algorithmic modeling are the three primary types into which software maintenance cost estimating approaches are divided in this research. Utilize statistical methods and mathematical formulas to compute maintenance costs in algorithmic modeling depending on fault density, complexity, and lines of code, among other variables. Rather of using algorithmic computations, non-algorithmic-based models rely on previous data, expert judgment as well as qualitative considerations. To calculate the cost of updating and maintaining software systems, knowledge-based models rely on expert knowledge. When there is little or no historical data available for a particular project, these models are frequently employed. Additionally, the study highlights the models that are most relevant to various project situations and domains, examines notable models within each category, and assesses the models’ attributes. Furthermore, the limits and difficulties related to software maintenance cost prediction are discussed in this study. This research examines and addresses the poor accuracy of current estimation models used in early maintenance phases, particularly when existing data is insufficient. A comprehensive comparative study about various estimation models, like knowledge-based and hybrid estimating techniques, is presented to bridge this research gap. Along with these new developments, it discusses how artificial intelligence is being included into estimating to make it more precise and flexible.