Performance Prediction of Configurable softwares using Machine learning approach
Tanuja Shailesh, Ashalatha Nayak, Devi Prasad · 2018
In the current software industry most of the complex softwares are configurable. Configurable software include different features that are considered essential for the functioning. Certain configurable features can have higher impact on system functional behaviour when compare to other features. A combination of different features selected result into a configuration space. There is a enormous increase in configuration space as the number of features increases. Each configuration in configuration space produces different system performance. Hence, there is a need to study the impact of different configuration on the system performance. Predictive models offer solutions to analyze system performance for a given configuration set. In this paper different machine learning techniques are compared and we propose a comparative results using WEKA tool. We propose a Neural network model with statistical techniques for predicting system performance for input configuration.