Training data selection for cross-project defect prediction

Abstract

Software defect prediction has been a popular research topic in recent years and is considered as a means for the optimization of quality assurance activities. Defect prediction can be done in a withinproject or a cross-project scenario. The within-project scenario produces results with a very high quality, but requires historic data of the project, which is often not available. For the cross-project prediction, the data availability is not an issue as data from other projects is readily available, e.g., in repositories like PROMISE. However, the quality of the defect prediction results is too low for practical use. Recent research showed that the selection of appropriate training data can improve the quality of cross-project defect predictions. In this paper, we propose distance-based strategies for the selection of training data based on distributional characteristics of the available data. We evaluate the proposed strategies in a large case study with 44 data sets obtained from 14 open source projects. Our results show that our training data selection strategy improves the achieved success rate of cross-project defect predictions significantly. However, the quality of the results still cannot compete with within-project defect prediction.

Software Defect Prediction Model Research for Network and Cloud Software Development

“…(1) Source selection algorithm for software project defect prediction Prediction of cross project software defects, between source and target attribute measure distribution project project there is a certain correlation [5] , so the maximum minimum value, and the mean and standard deviation of 4 data distribution data to define the feature vector of the C project. To help achieve the feature similarity sort source and target selection of software project project the distribution of software module is a module attribute, expressed as C = {c max , c min , c mean , c std }, the distribution characteristics of each project data attribute values can be used in c(F m (S))to represent the feature vector of the S project, which can be defined as follows:…”

Combing Data Filter and Data Sampling for Cross-Company Defect Prediction: An Empricial Study

“…The first one is to apply the data filtering method to find the best suitable training data (e.g., [8,9,12,14]). For example, Turhan et al [8] proposed a nearest neighbor (NN) filter to select cross-company data.…”

A Ranking-Oriented Approach to Cross-Project Software Defect Prediction: An Empirical Study

“…After Briand et al made an early attempt to validate the applicability of CPDP [14], many researchers in this field have tried to improve the performance of CPDP models using different techniques such as data mining and machine learning. Fortunately, recent studies have shown that it is indeed a feasible method for defect prediction in software projects with different sizes [13,[15][16][17][18][19][20]. Due to space limitations of this paper, for more details about CPDP approaches, please refer to the latest surveys [6,7].…”