Sigma Worksheet: Interactive Construction of OLAP Queries

The new generation of cloud data warehouses (CDWs) brings large amounts of data and compute power closer to users in enterprises. The ability to directly access the warehouse data, interactively analyze and explore it at scale can empower users to improve their decision making cycles. However, existing tools for analyzing data in CDWs are either limited in ad-hoc transformations or difficult to use for business users, the largest user segment in enterprises. Here we introduce Sigma Worksheet, a new interactive system that enables users to easily perform ad-hoc visual analysis of data in CDWs at scale. For this, Sigma Worksheet provides an accessible spreadsheet-like interface for data analysis through direct manipulation. Sigma Worksheet dynamically constructs matching SQL queries from user interactions on this familiar interface, building on the versatility and expressivity of SQL. Sigma Worksheet executes constructed queries directly on CDWs, leveraging the superior characteristics of the new generation CDWs, including scalability. To evaluate Sigma Worksheet, we first demonstrate its expressivity through two real life use cases, cohort analysis and sessionization. We then measure the performance of the Worksheet generated queries with a set of experiments using the TPC-H benchmark. Results show the performance of our compiled SQL queries is comparable to that of the reference queries of the benchmark. Finally, to assess the usefulness of Sigma Worksheet in deployment, we elicit feedback through a 100-person survey followed by a semi-structured interview study with 70 participants. We find that Sigma Worksheet is easier to use and learn, improving the productivity of users. Our findings also suggest Sigma Worksheet can further improve user experience by providing guidance to users at various steps of data analysis.