Search academic texts

Information × Registration Number 0222U002207, 0121U109527 , R & D reports Title Development of information technology for software reliability assessment and forecasting by machine learning methods popup.stage_title Head Yakovyna Vitaliy S., Доктор технічних наук Registration Date 10-02-2022 Organization Lviv Polytechnic National University popup.description2 The research used a dataset obtained by merging KS1, KS2, PC1, CM1, and JM1 datasets from the PROMISE Software Engineering repository, which contained data on testing software modules and 21 code metrics. Boruta, Stepwise selection, Exhaustive Feature Selection, Random Forest Importance, LightGBM Importance, Genetic Algorithms, Principal Component Analysis, Xverse python methods were used to select the most important features that affect the quality of program code. Based on the voting of the results of feature selection, a static (deterministic) software reliability model is built using logistic regression, which establishes the relationship between the probability of a defect presence in the software module and the metrics of its code. The increase in the accuracy of forecasting defective software modules in the case of using the developed model (compared to the initial data set) ranged from 10% to 21%. It has been shown that the greatest contribution to the probability that a software module contains one or more defects is made by McCabe's cyclomatic complexity, the number of branches in the program, and the Halstead’s number of operators and operands.  Product Description popup.authors Izonin Ivan V Boyko Natalya I Kryvenchuk Yurii P Melnykova Natalya I Symets Ivan I Uhrynovskyi Bohdan V Fechan Andriy V Shakhovska Natalya B popup.nrat_date 2022-03-09 Close