Pocket Estimator -- A Commercial Solution to Provide Free Parametric Software Estimation Combining an Expert and a Learning Algorithm

摘要

Pocket Estimator is a cloud-based framework to combine an expert weighted estimation algorithm with several learning algorithms for high level, parametric software effort estimation. Main goal of our framework is to create a huge estimation dataset of software implementation projects. This database will be built over the next 2 years and should be used for further scientific research in learning and adjusted effort estimation. We have implemented a k-nearest-neighbor and an expert weighted estimation algorithm. This paper presents our framework and describes the interaction of the parametric software estimation algorithms.