An enhancement of software metrics as failure predictors

Software has become a part of almost every device around us, from mobile phones to medical systems; Failure within such devices can have fatal consequences. Thus, the need for eliminating these failures is inevitable. This study attempts to find a correlation - using Pearson and Spearman correlation coefficients - between code failures and software metrics with the goal of enhancing failure predictors. The results show statistically significant correlations between software metrics and code failures. This study is unique in testing certain combinations of metrics that are highly correlated with code failures.