This trial version of Kaskada's time-based feature engine unleashes instant iteration on what feature definitions are and when training examples are observed -- with data-dependent, entity-specific computations.

At a high-level Kaskada is a feature engine that allows you to connect directly to your event-based data and defer decision-making. Instead of any one person having to decide when to take snapshots or what data to collect, connect directly to your event-based data where you track everything.

Kaskada gives you a way to express what you want to compute and what points in time are important. This gives domain experts a way to compute instantly and iterate on features with historical data and deploy those features, as code, seamlessly to production for online feature scoring.

Use this product version to try out our instant iterative data science experience where you can iterate thousands of times in weeks and not months to create production-ready features without leakage.

Kaskada is time and entity aware; our abstractions allow you to go from more than 63 pages of code that manipulate event-based data to just two pages of readable, sharable, and composable queries.

Join your data without leakage:

• Connect to all your event-based data sources
• Group and re-group data to associate with Entities
• Create new tables and views as needed
• Perform temporally accurate joins across Entities

• Love Jupyter for experiments? Use our ipython extension.
• Leverage notebooks for experimentation
• Build up features and time selection during exploration
• Change time selection to affect all features without needing to make data pipeline changes

• Enable sharing, preserve privacy, prevent data leakage
• Share feature definitions, not just results
• Easily compute the same features over different datasets