Dataset For Software Defect Predictions

SOFTLAB: dataset were from a Turkish software company which develops embedded controllers for home appliances [1]
RELINK: Dataset with has 26 static code
features and contains three projects (Apache HTTP
Server (Apache), OpenIntents Safe (Safe), and ZXing) [2]
NASA: Dataset from e C-based and are
extracted from a software system which is made of a set
of static code features [3].
4. AEEEM: dataset was compiled by
Marco et al. [4], and contains five open source projects
with 5,371 data points. The projects include: Apache
Lucene (LC), Equinox (EQ), Eclipse JDT Core (JDT),
Eclipse PDE UI (PDE) and Mylyn (ML).

AEEEM software defect datasets which has a format of ARFF was collected by D’Ambros et al. [1]
[1] M. D’Ambros, M. Lanza, and R. Robbes, “Evaluating defect prediction approaches: A benchmark and an extensive comparison,”Empirical Softw. Engg., vol. 17, no. 4-5, pp. 531–577, Aug. 2012.
[3] ] B. Ghotra, S. McIntosh, and A. E. Hassan, “Revisiting the impact of
classification techniques on the performance of defect prediction
models,” in 2015 IEEE/ACM 37th IEEE International Conference on
Software Engineering, vol. 1. IEEE, 2015, pp. 789–800.
[4] M. D’Ambros, M. Lanza, and R. Robbes, “Evaluating defect prediction approaches: a benchmark and an extensive comparison,”
Empirical Software Engineering, vol. 17, no. 4, pp. 531–577, 2012