DESIGN CONCEPTION OF INFORMATION DASHBOARDS FOR MASSIVE DATA VISUALIZATION

Abstract

Dashboards as a data visualization tool, are business reporting tools that aggregate all data in a single screenshot. Dashboards are links to data files, application program interfaces (APIs) and metadata to provide users major metrics and key performance indicators (KPI) about systems' processes or a business entity. Hence, Dashboards extract, aggregate highlight and communicate high-level information to infer anomalies, prospects, issues and trends. This study proposed Wander join technique to provide data aggregation for fast interactive queries for data visualization with minimal latency. The importance sampling approach of the Wander join algorithm was used to support joins for data convergence and augment latency in common visualization queries. This enabled a uniform convergence rate for all displayed data aggregation categories. 30,000 samples out of 120 million records of flights arrival and departure at an airport were drawn with a convergence of 0.05 relative error with near-zero latency as the aggregation occurs. With adjustable weights, the Wander Join algorithm allows groups of interest to be sampled often, while groups of less interest can be weighted to be sampled less often if the weight is set to 0.