Dataset and Bayesian Net Model for Fenton et al “Project Data
Incorporating Qualitative Factors for Improved Software Defect
Prediction” (a paper presented at PROMISE 2007).
There are two datasets associated with the paper. The first is an exel file
containing qualitative and quantitative data about 31 projects
completed in a consumer electronics company. The columns of the
spreadsheet are the projects and the rows are the attributes measured
for each project. The first column names the attributes explicitly.
There is a mixture of qualitative attributes (these are measured on a 5
point ranked scale VL, L, M, H, VH) and quantitative attributes whose
scale is stated.
The other ‘dataset’ is the Bayesian net model
(in cmp file format) that was the subject of the validation study. This
model was built over a number of years involving a several
organisations. To view and run the model you will need to install the
AgenaRisk software, which can be downloaded for free (for one month)
The desc file describing the data is here.