Y. Singh, B. Goel, and P. Chandra (India)
Corrective maintenance, SFFANNs, Software Measures.
This paper presents the application of neural networks in
predicting corrective maintenance time. The paper aims to
establish the viability of the usage of feedforward
artificial neural network for predicting the time needed to
correct errors associated with changes made to the
software during maintenance. For this purpose a neural
model using four software measures is proposed. The
software measures used are complexity measures like
cyclomatic complexity (ACC), readability of source code
(RSC), documentation quality (DOQ) and
understandability of software (UOS). The neural network
used is Sigmoidal Feedforward Artificial Neural Network.
It is found that the neural network of the type used can act
as an efficient predictor of corrective maintenance time.