Operational Software Reliability Prediction by Random Forest Based on Development Project Data with Qualitative Variables  
Author Y. Arai


Co-Author(s) M. Kimura


Abstract This article discusses a method for finding the software development project in which the developed software may cause at least one software failure in its operational phase. For this, we analyze the data sets which were obtained as the results of questionnaire from the real software development companies. In particular, the data sets consist of not quantitative variables but qualitative ones. As a result of the actual data analysis, we found that the Random Forest showed little bit better performance against the traditional multiple regression analysis with dummy variables.


Keywords Software reliability, qualitative variable, machine learning, Random Forest
    Article #:  21109
Proceedings of the 21st ISSAT International Conference on Reliability and Quality in Design
August 6-8, 2015 - Philadelphia, Pennsylvia, U.S.A.