June 28, 2023

Snowpark and Sigma Open Machine Learning to Business Teams

Data science models built in Snowflake are now available to the business user through Sigma's BI interface.

For data science professionals, open-source workflows typically involve a mess of data exports, external compute, and stale output files. With Snowpark, data science teams can bring Python and other languages for data science and machine learning (ML) directly to Snowflake, where an organization’s data lives. In doing so, ML models can run inside each Snowflake end-customer’s account where their data is secure and governed. Since released into General Availability, Snowpark for Python has continued to grow in adoption among data teams looking to effectively run data science and machine learning at scale within their organization.

For the data science professionals building the ML models, Snowpark allows programmers to break out of desktop tools, or siloed development in platforms other teams could not easily access. A key differentiator of Snowpark is that ML teams have the option to continue development from their editor of choice and push down processing to Snowflake or work directly from a Snowflake Python Worksheet. When the models are ready, developers can make the ML model available inside Snowflake for other users to score their data and gain new insights. For the data consumers and business analysts, this means that they can use ML models in the same place that they are getting their Snowflake data. Beyond the performance and security aspects that come with bringing the processing to the data, organizations can now count on the speed, concurrency, and extensibility to develop and run data applications, models, and pipelines where data lives.

With the Sigma on Snowpark workflow, consumers and business analysts can access data science insights alongside their Snowflake data, eliminating silos and promoting collaboration.

How does this relate to Sigma? As Snowpark breaks down barriers within your data team, the Sigma platform will make sure the front line of your business benefits as well, with completely managed Snowflake push-through functionality. Sigma’s startup-tested, enterprise-approved data exploration platform brings the familiarity of the spreadsheet alongside the power of the Snowflake data cloud. Every new Snowpark model your business develops can be registered as a governed function that your line-of-business users will be able to call in a spreadsheet interface, just like the spreadsheet functions they’re used to.

Using time-series functions, a company can forecast future orders, as well as enhance past data with context that shows a live impact to the prediction.

Snowpark and Sigma have revolutionized the landscape for data science professionals and data consumers alike. With Snowpark's ability to execute Python, Javascript, and Scala directly on the secure Snowflake platform, data scientists can break free from cumbersome workflows and consolidate their operations within the unified Snowflake data cloud. This seamless integration enables data consumers and business analysts to access data science insights alongside their Snowflake data, eliminating silos and promoting collaboration. By combining the power of Snowpark's processing capabilities with Sigma's user-friendly interface, organizations can unlock limitless possibilities for developing and running data applications, models, and pipelines. With Snowpark and Sigma, the barriers within your data team dissolve, empowering your entire organization to leverage the full potential of your data.

And that means functionally limitless possibilities.

Sigma on Snowpark Examples

Take a look at these sample functions at work

Sigma & Snowpark Classification: Run new or hypothetical customers through a governed loyalty program model.

Sigma & Snowpark Forecasting: Automatically assess historical trends and project those into the future, using Input Tables to augment the ML predictions on the fly.

Sigma & Snowpark Clustering: Use ML techniques to automatically categorize your data into natural clusters, whether those are regions, stores, or customers.