Source code survival with the Kaplan Meier

The presence of dead code may affect the comprehensibility, the readability, and the size of source code so increasing the effort and the cost for maintenance. The elimination of dead code needs a huge cost and effort for recognizing and eliminating code that is not effectively used. The goal of this work consists in defining an approach based on the Kaplan Meier estimator to analyze dead code. The validity of the approach has been preliminarily assessed on a case study constituted of fifty-eight versions of five open source software systems implemented in Java. The results suggested that two out of the five systems where implemented avoiding as much as possible the introduction of dead code.