New Ideas and Emerging Research: Evaluating Prediction System Accuracy
Source: Brunel University
An important feature of any engineering discipline is the ability to make timely and accurate predictions, and in order to do so people need prediction systems. Moreover the authors need to evaluate such models or prediction systems. Software engineering is no exception. However, judging predictive accuracy is a subtle task and their failure to do so appropriately is a contributor to the present situation of a lack of conclusion stability across studies and inconclusive results from systematic reviews.