Software Defect Prediction by Online Learning Considering Defect Overlooking

TL;DR

This paper investigates how online learning-based defect prediction models can overlook defects due to negative predictions and erroneous test results, highlighting the negative impact on prediction accuracy.

Contribution

It identifies the issue of defect overlooking caused by negative predictions in online learning models and demonstrates its negative effect on accuracy through experiments.

Findings

Negative predictions can lead to defect oversight.

Erroneous test results negatively impact online learning accuracy.

Online learning models are sensitive to misclassified data.

Abstract

Building defect prediction models based on online learning can enhance prediction accuracy. It continuously rebuilds a new prediction model when adding a new data point. However, predicting a module as "non-defective" (i.e., negative prediction) can result in fewer test cases for such modules. Therefore, defects can be overlooked during testing, even when the module is defective. The erroneous test results are used as learning data by online learning, which could negatively affect prediction accuracy. In our experiment, we demonstrate this negative influence on prediction accuracy.