Study presented in this paper is based on analysis of network bandwidth usage and using it as baseline to predict and improvise future network bandwidth usage. Network bandwidth plays an important role in evaluating any application’s performance running on a computer. Although it is not an application dependent parameter, however if affects overall performance of all the applications. Network bandwidth is finite and is limited by the laws of physics as well as technology limitations. Also, it is not free. For any organization, the cost of network bandwidth adds as a major and continual expense in overall infrastructure costs. Thus it becomes very important to continually monitor, analyze and improvise network bandwidth usage pattern. Historic data of network bandwidth usage is captured and used in this study. This historic data after some transformation defines the baseline, which is then absorbed by predictive analytics module. In this study we combine business intelligence along with bandwidth analytics. Business intelligence technique called as Stacking of predictive models is used to optimize the network usage patterns and thus improvise the overall usage of network bandwidth. An exhaustive research is also done while choosing the best combination of algorithms for stacking.