DataScope: Interactive visual exploratory dashboards for large multidimensional data

摘要

DataScope is a web-based tool for generating interactive visual dashboards on large scale multidimensional datasets. Users create these dashboards using a high-level declarative grammar, and use them to explore data and create cohorts for downstream analysis. We describe DataScope's architecture, design considerations and provide an overview of the system. We highlight some of DataScope's features that were useful in the case studies using datasets from cancer registries and co-clinical trials. In benchmarks DataScope is able to perform sub-second queries on data sizes ranging from thousand to million records.