An Analysis of Accuracy and Learning in Software Project Estimating

This paper presents a study into the accuracy of different dimensions of IT project estimating: schedule, budget & effort. The study is based on a dataset of 171 projects that have been collected at the IT-department of a large Dutch multinational company. The paper also analyses whether there is any learning (improvement) effect over time. Our results show that there is no relation between accuracy of budget, schedule and effort in the analyzed organization. Besides, they show that over time there is no change of the accuracy effectiveness and efficiency. The paper reflects and provides recommendations on how to improve the learning from historical estimates.