Predicting software bugs using ARIMA model

The number of software products available in market is increasing rapidly. Many a time, multiple companies develop software products of similar functionalities. Thus the competition among those owning companies is becoming tougher every day. Moreover, there are many crucial programs whose results should be always accurate without fail. As a consequence of such challenges, tackling software bugs issues efficiently is an important and essential task for the owning software companies. Therefore, predicting bugs and finding ways to address these at the earliest has become an important factor for sustainability in the software market. This paper proposes software bug predication models using Autoregressive Moving Average Model (ARIMA) based on Box-Jenkins Methodology, which depends on Autoregressive models (AR) with Moving Average (MA). The inputs to our models are the information extracted from the past bug repositories. We have verified our models using datasets of Eclipse [16] and Mozilla [17].