skip to main content

Ovum view


Arcadia Data is one of a rapidly growing panoply of vendors embedding machine learning (ML) into their solutions. Already using ML to define the underlying OLAP views (query accelerators known as Analytic Views), the new release is taking the next step toward harnessing the capability to recommend how to visualize the data. It's the next logical step toward delivering the guided analytics experience that self-service analytics tools have introduced.

Embracing machine learning to drive the user experience

Machine learning is steadily becoming a checkbox item for next-generation analytic tools. This has been especially pronounced in the cloud, where services such as Amazon QuickSight and IBM Watson Analytics have provided aids that guide users on recommended data sets, narratives, and visualizations. This has spread to the on-premises world where, for instance, Tableau's long-term public roadmap encompasses recommendations based on the user's query history, natural language query, and automated discovery for hidden insights. Arcadia Data was already using machine learning under the hood to generate query-accelerating Analytic Views based on data utilization, with a feature known as Smart Acceleration. With the new release, Arcadia Data also makes recommendations to end users on how to visualize the results of their queries via Instant Visuals. This is clearly not the endgame for Arcadia Data, as we expect it to keep adding new features to further guide the user. We believe the next step would be to make recommendations regarding the use of data sets relevant to the query stream.

Another key enhancement to the new release is extensibility to its support for complex data types. Arcadia Data already can import and query relatively simple JSON files, such as weblogs – but so can most of its rivals. The latest version ups the ante with more complex data types such as arrays, maps, and structs, where the data structures are sufficiently intricate to be beyond the capabilities of most BI tools. Another differentiator with the latest version is that the tool avoids the commonplace approach of SQL-oriented BI tools for flattening complex data types into wide columns, or flattening data sets with sparse data through populating redundant rows. Instead, Arcadia Data represents the hierarchy in its UI so analysts can see the original structure and relationships in the nested data. Another extensibility enhancement is being the first BI vendor to support KSQL, enabling Arcadia Data users to visualize real-time data that is being streamed via Kafka.

Arcadia Data has competed with self-service BI tools with its ability to directly query Hadoop; the new capabilities make it more competitive with the user experience.


Further reading

On the Radar: Arcadia Data unifies visual analytics and BI on Hadoop, IT0014-003266 (May 2017)

"Splice Machine's native Spark support adds path for data scientists," INT002-000089 (April 2018)

"Self-service comes to machine learning," INT002-000081 (February 2018)


Tony Baer, Principal Analyst, Information Management

[email protected]