New auto-aggregation features
Pecan’s feature-engineering process automatically creates new features so you can extract valuable information, which would otherwise be overlooked, from your dataset.
We recently added two new auto-aggregation features, which have been shown to improve model performance by 10-30%:
Mode – Extracts the value that occurs most frequently in a user’s history.
Median - Extracts the median from a range of values (particularly important when a column has extreme outliers).
PostgreSQL support
Pecan can now connect to data stored in PostgreSQL, and use it to train models and generate predictions.
Here are Pecan's currently supported connectors (with more coming soon):
These are the fields you’ll complete when configuring a PostgreSQL connection.
Exporting predictions to Redshift
Pecan now supports exporting predictions to an Amazon Redshift database. To enable this functionality, see How to configure an Amazon Redshift connection.