Title :
A study of applying ARIMA and SVM model to software reliability prediction
Author_Institution :
Dept. of Appl. Inf., Fo Guang Univ., Yilan, Taiwan
Abstract :
For more than three decades, Box and Jenkins´ Auto-Regressive Integrated Moving Average (ARIMA) technique has been one of the most widely used linear models in time series forecasting. However, it is well documented that many software failure observations are nonlinear and ARIMA is a general univariate model developed based on the assumption that the time series data being predicted are linear. Therefore, in this study, the utilization of Support Vector Machine (SVM) as a nonlinear model and ARIMA as a linear model are integrated in software reliability forecasting. Experiments on real-world data set validate the effectiveness of the hybrid model. These results also show that the proposed methodology can be a more effective way in order to combine linear and nonlinear models together than traditional methodologies. Therefore, it can significantly improve the prediction performance and can be applied as an appropriate alternative approach for software reliability forecasting field, especially when higher prediction performance is needed.
Keywords :
autoregressive moving average processes; software reliability; support vector machines; time series; ARIMA; SVM model; auto-regressive integrated moving average technique; software reliability prediction; support vector machine; time series forecasting; Artificial neural networks; Data models; Forecasting; Predictive models; Software; Software reliability; Support vector machines; Autoregression Integrated Moving Average (ARIMA); Software Reliability; Support Vector Machine (SVM);
Conference_Titel :
Uncertainty Reasoning and Knowledge Engineering (URKE), 2011 International Conference on
Conference_Location :
Bali
Print_ISBN :
978-1-4244-9985-4
Electronic_ISBN :
978-1-4244-9984-7
DOI :
10.1109/URKE.2011.6007794