NFA provides a novel intelligent method for estimation. Our method is applicable to various estimation problems and it increases the accuracy of estimation. The estimation system comprises of the following components:
• Pre-processing neuro-fuzzy inference system (PNFIS) used to resolve the effect of dependencies among contributing factors (RF) of the estimation problem, and to produce adjusted rating values (ARF) for these factors,
• Neuro-fuzzy bank (NFB) used to calibrate the contributing factors by mapping the adjusted rating values for the contributing factors to generate their corresponding numerical parameter values (FM),
• Module that applies an algorithmic model relevant to the nature of the estimation problem to produce one or more output metrics (Mo).

large figure

NFA relates generally to the above novel and inventive estimation model and framework that makes improved use of both historical project data and available expert knowledge, by uniquely combining certain aspects of relatively newer estimation techniques (e.g., neural networks and fuzzy logic) with certain aspects of more conventional software estimation models (e.g., COCOMO, function point, SEER-SEM), to produce more accurate estimation results.
Our model is inherently independent of the choice of algorithmic models and nature of the estimation problems. It has learning and adaptation ability, integrates capability of expert knowledge, project data and parametric algorithmic models, and provides robustness to imprecise and uncertain inputs. It also has good interpretability and high accuracy.
Our validated results show that applying our method to to COCOMO, function point and SEER-SEM, after training and learning, improves the estimate by approximately 20%.