Seasonal influenza is a contagious respiratory illness that can cause various complications, worsen chronic illnesses, and sometimes lead to deaths. During 2009 H1N1 flu pandemic, up to 203,000 deaths occurred worldwide. Early detection and prediction of disease outbreak is critical because it can provide more time to prepare a response and significantly reduce the impact caused by a pandemic. The traditional influenza surveillance system by Centers for Disease Control and Prevention (CDC) collects U.S. Influenza-Like Illness related physicians visits data from sentinel practices and provides a retrospective analysis delayed by two weeks. Google Flu Trends proposed a method that uses online search queries data to estimate current (real-time) influenza activity. Here we present a system that (1) predicts future influenza activities, (2) provides more accurate real-time assessment than before, and (3) combines real-time big social media data streams and CDC historical datasets for predictive models to accomplish accurate predictions. Although retrospective analysis and observations are important, prediction of future flu levels can represent a big leap because such predictions provide actionable insights for public health that can be used for planning, resource allocation, treatments and prevention. Thus, compared to previous work, our work represents an advancement in accuracy of assessments, prediction of future flu activity accurately and an ability to combine big social data and observed CDC data to build predictive models.