A Proposal for Analysis and Prediction for Software Projects using Collaborative Filtering, In-Process Measurements and a Benchmarks Database
Yoshiki Mitani, Nahomi Kikuchi, Tomoko Matsumura, Naoki Ohsugi, Akito Monden, Yoshiki Higo, Katsuro Inoue, Mike Barker, Kenichi Matsumoto · OUKA (Osaka University Knowledge Archive) (Osaka University) · 2006
Abstract. This paper proposes a new method for developing predictions and estimates for ongoing projects by comparing in-process measurements of the current project with benchmark data from previous projects. The method uses collaborative filtering to identify groups of similar projects in the benchmark database and then to develop predictions and estimates based on the in-process measurements of the current project and the comparison data from the similar projects. The authors base this proposal on experiments with multidimensional in-process project measurement in a middle-scale multi-vendor development that lacked transparency in its processes. The authors ' measurement trial verified the usefulness of the measurement methods, especially in project management, as reported in the paper.