Artificial Intelligence Techniques - Novel Approaches & Practical Applications |
Foundation of Computer Science USA |
AIT - Number 3 |
None 2011 |
Authors: Roheet Bhatnagar, Mrinal Kanti Ghose, Vandana Bhattacharjee |
b0f03a86-cc2e-4f03-934b-35447f79484e |
Roheet Bhatnagar, Mrinal Kanti Ghose, Vandana Bhattacharjee . A Novel Approach to the Early Stage Software Development Effort Estimations using Neural Network Models: A Case Study. Artificial Intelligence Techniques - Novel Approaches & Practical Applications. AIT, 3 (None 2011), 8-11.
Software development is a team activity which has got well defined stages. At every stage we have a milestone and a milestone needs to be achieved in order to move to the next level. On time and within budget delivery is the ultimate goal of software projects. Software project management is the set of processes and guidelines which help us in achieving the on time & within budget product delivery. Lot of studies has been carried out on software development effort estimations which forms a key part of the software project management. Many models have been proposed over the years for cost estimations but it is still a subject of constant research due to the ever changing nature of software development. This paper describes a novel approach towards the effort estimations at the early stages of software development life cycle (SDLC). A dataset has been created based on the Entity Relationship Diagrams (ERD’s) developed by different engineering students as part of their Major Project’s spreading different batches & year. In this paper three different NN models are used and their results are compared based on the standard evaluation criteria’s such as MMRE, MRE, BRE and Pred(20).