A data-driven model for software reliability prediction

In the actual software development, failure data is rarely pure linear or nonlinear. It is usually formed by the linear and nonlinear patterns at the same time. These models can be divided into two main categories: analytical model and data-driven model. Analytical SRMs are proposed based on underlying assumptions about the nature of software faults, the stochastic behavior of the software processes and the development environments. On the contrary, the so-called data-driven models, borrowing heavily from artificial intelligence techniques, rely directly on the collected data describing input and output characteristics. Compared to analytical SRMs, data-driven models have much less unpractical assumptions and are much abler to make abstractions and generalizations of the software failure process. It has been recognized that the auto regression integrated moving average (ARIMA) and the support vector machine (SVM) perform fairly well in predicting linear and nonlinear time series data. Therefore, we propose a hybrid approach to software reliability forecasting using both ARIMA and SVM models.