Using code change types in an analogy-based classifier for short-term defect prediction

Authors:
Josée Tassé
Affiliations:
University of New Brunswick, Saint John, New Brunswick, Canada
Venue:
Proceedings of the 9th International Conference on Predictive Models in Software Engineering
Year:
2013

Citing 5
Cited 0

Analogy-Based Practical Classification Rules for Software Quality Estimation

Empirical Software Engineering
Predicting Faults from Cached History

ICSE '07 Proceedings of the 29th international conference on Software Engineering
Classifying Software Changes: Clean or Buggy?

IEEE Transactions on Software Engineering
Benchmarking Classification Models for Software Defect Prediction: A Proposed Framework and Novel Findings

IEEE Transactions on Software Engineering
Predicting faults using the complexity of code changes

ICSE '09 Proceedings of the 31st International Conference on Software Engineering

Quantified Score

Hi-index	0.00

Visualization

Abstract

Current approaches for defect prediction usually analyze files (or modules) and their development as work is done on a given release, to predict post-release defects. What is missing is an approach for predicting bugs to be detected in a more short-term interval, even within the development of a particular version. In this paper, we propose a defect predictor that looks into change bursts in a given file, analyzing the number of changes and their types, and then predict whether the file is likely to have a bug found within the next 3 months after that change burst. An analogy-based classifier is used for this task: the prediction is made based on comparisons with similar change bursts that occurred in other files. New metrics are described to capture the change type of a file (e.g., small local change, massive change all in one place, multiple changes scattered throughout the file).