Article Preview
TopIntroduction
Accurate estimation of software projects plays crucial role in the positive outcome of the project. The estimation further impacts staffing, costing, planning, resource planning, milestone planning and such activities. There are very well-established estimation methods for software projects for matured project execution methodologies such as waterfall model, Iterative model. However modern software projects pre-dominantly use agile methodologies that have shorter sprints, continuous and iterative delivery.
Agile execution model aims to deliver the software in smaller increments leading to faster delivery cycles. Scrum, extreme programming (XP), feature driven development, DSDM (Dynamic System Development Method), adaptive software development are the main methods that follow agile methodology. Four main manifestos of agile projects are “Individuals and interactions over processes and tools.”, “Working software over comprehensive documentation”, “Customer collaboration over contract negotiation” and “Responding to change over following a plan”. Due to the flexible and dynamic nature of attributes associated with agile characteristics, traditional estimation models cannot be used for agile estimation.
Story point estimation and use case point estimation are normally used for agile projects. Agile projects are more responsive to change and promote collaborative and iterative deliveries (Highsmith, 2001). “User story” is the primary unit of work for developers in agile project. Developers define tasks, timelines and dependencies for user stories and user stories are prioritized based on their business importance.
In this paper we discuss a novel estimation method, “Normalized sprint estimation” for agile projects that factors in the core agile characteristics during estimation.
Paper Organization
In the remaining portions of the introduction section we will look at state of the art methods and the challenges and gaps with state of the art agile effort estimation methods. We will discuss the complete details of the “Normalized sprint estimation” in the “Method” section. In “Results” section we will look at the MRE, MMRE and Pred (0.25), Pred (0.3) values for 14 sprints taken from three different digital project. Finally, we will discuss the results, threats to validity and future scope of improvements in “discussion” section.