A Risk Management Model Based On Neural Network Algorithms

Software Risk Management is always a hot research field in software project management industry. Study in this field could push forward the quality of enterprise's project management. Also improve software's availability, external image and company's software process capability. But the true is that over half of software could not get the previous quality level or goal in every year. We waste not only money but also business market, client, prestige and get into trouble with the law if even worse. So, enhance the level of risk management, reduce the development risk is your company's survival plan. From Doctor Boehm bring the theory of Software Risk Management up now on, appear many classic models. But most of all only stay in theoretical stage. The others either too difficult to operate or take practical situation of Chinese company into account. We can not get desired effects if copy these Ready-Made model. Based on this purpose, this paper design a new model based Boehm and CRM software risk management model. It using Neural Network as its risk forecast and identification, also amend these two classic models. This new model chose continuous management method, hope identify and control risk earlier and improve software's quality step by step. At the end, we use an actual project which used the model as empirical data, compared with software quality without the model. Experimental result show that under the model, software quality improved obviously. Meanwhile, this can prove this new model do well in controlling software Risk.