Welcome to the Empirical Software Engineering’s special issue on predictive models in software engineering. The goal of such methods is repeatable, refutable (and possibly improvable) results in software engineering.

Many of the recent papers in SE literature are based on data from on-line repositories such as http://promisedata.googlecode.com. This introduces a kind of selection in the kinds of papers published at this venue. Our first paper pushes past that bias to explore a very rich time-based data set. In “Predicting the Flow of Defect Correction Effort using a Bayesian Network Model”, Schulz et al. use a Bayes net to explore the effects of removing defects at different stages of the software lifecycle. Their work shows how to calibrate general models to the particulars of a company’s local particulars.

Our next paper “The Limited Impact of Individual Developer Data on Software Defect Prediction” by Bell et concludes there is no added value to reasoning on some aspects of social aspects of programmer teams working on a code. This is a timely counterpoint to other research that eschews code measures for other approaches based only on social metrics.

Our last paper explores the complicated issue of parameter tuning. In “Using Tabu Search to Configure Support Vector Regression for Effort Estimation”, Corazza et al. offers automated guidance for setting the parameters that control a learner. This is a matter of critical importance since even the best learner can perform poorly if its operator uses the wrong settings.

A special issue like this is only possible due to the hard work of a dedicated set of authors are reviewers. We would like to express our gratitude to all authors who submitted their papers this special issue. We would also like to thank our reviewers for their meticulous evaluation of the submissions. The success of special issues such as this one largely stands on their shoulders.