May 2022 Summaries

Superset's mixed time-series visualization feature, first introduced in version 1.2, allows for the overlaying of multiple time-series charts within a single chart container, providing an enhanced platform for comparative data analysis. This capability, leveraging Apache ECharts instead of the older NVD3, enables users to combine different types of charts, like line and bar graphs, to explore relationships and trends in temporal data more effectively. Mixed time-series charts are particularly useful for comparing subsets of data, such as daily and weekly active users, against a total population, or for visualizing rolling means to smooth out data fluctuations. The shared axes and charting space facilitate clearer comparisons, making it easier to identify patterns, trends, and anomalies. Superset achieves this by running separate SQL queries for each data series, merged by a common datetime range and grain, allowing users to craft custom visualizations that can be further tailored through chart customization options.

Apache Superset 1.5 introduces several enhancements and new features aimed at improving user experience and performance, particularly in dashboard management and SQL customization. The update provides faster loading times for complex dashboards, a cleaner design for dashboard filters, and improved interdependencies between filters. Users can now write custom SQL expressions directly in charts without saving them to the dataset, thanks to a new Custom SQL tab in the Column control. Additionally, Superset 1.5 offers a better caching experience by storing dashboard and Explore states in the Superset metadata store, reducing issues related to long URLs. The release also includes numerous bug fixes and updates to existing database support, such as enhancements for DuckDB and Azure Data Explorer. Superset 1.5 is available for upgrade via multiple methods, including a Python package and Docker image, with further details and changelogs accessible through the Superset GitHub repository.

Jinja templating is a powerful tool for enhancing the interactivity and flexibility of dashboards in Superset and Preset, enabling users to create dynamic SQL queries without modifying the underlying data queries. Jinja, a Python templating engine, allows the integration of programmatic capabilities into SQL, thereby resolving SQL's limitations by supporting loops, variables, and context injections. This facilitates the creation of more interactive dashboards by enabling dynamic filtering capabilities, such as filtering based on user attributes or URL parameters. Superset's evolution included native filters, but Jinja allows more advanced and flexible query manipulations, including dynamic subqueries and the use of database-specific functions, thus increasing performance and user control. The process involves creating virtual datasets with Jinja templates, which can be adapted in dashboards to apply intricate filters and conditions, though care must be taken as extensive use of Jinja could complicate data lineage and trouble-shooting. Ultimately, Jinja empowers users to craft sophisticated, personalized visualizations while relying on Superset's fundamental strengths in data exploration and visualization.