This guide is designed to help you dive deeper into the Predictive Query Language (PQL), a SQL-like syntax that powers Kumo’s predictive modeling tasks. This reference will provide you with a more granular look at the commands and operators available in PQL.

The reference is organized into three main sections:

  • Primary Commands

    This section introduces the core building blocks of PQL, including commands such as ASSUMING, RANK, FOR EACH, PREDICT, and WHERE. These commands allow you to define the data context, specify your prediction targets, and structure your query for model training.

  • Aggregation Operators

    Here, you’ll find operators that perform aggregate computations on your data. Functions like AVG, COUNT, COUNT_DISTINCT, FIRST, LAST, LIST_DISTINCT, MAX, MIN, and SUM help you create summary statistics within your predictive queries.

  • Boolean Operators

    This section covers the logical and comparison operators used to filter and refine your queries. You’ll learn about operators such as AND, OR, NOT, along with pattern matching functions like CONTAINS, LIKE, NOT_LIKE, STARTS_WITH, and ENDS_WITH. These tools are essential for crafting complex conditions that accurately target your data.

Happy querying!

ASSUMING