Streaming insights are designed for near real-time data processing in Data Cloud. They perform aggregations and calculations on data as it is ingested from sources like the Mobile SDK and Interaction SDK. This allows for low-latency segmentation and activation based on user behavior within defined time windows, such as the 15-minute window specified in the question. Batch transforms and calculated insights operate on schedules (hourly or daily) and are not suitable for this real-time requirement.

A. Batch transform: This processes data in scheduled batches, which does not meet the near real-time, 15-minute window aggregation requirement.

B. Calculated insight: This runs complex SQL queries on a schedule (typically hourly or daily) over the entire dataset, not on incoming data streams.

D. Formula fields: This is a feature within standard Salesforce objects for row-level calculations, not a Data Cloud feature for streaming data aggregation.

1. Salesforce Help, Data Cloud, "Streaming Insights and Data Actions". This document specifies, "Use streaming insights to create data model object records that are based on events that are occurring in a stream... For example, you can calculate a customer’s propensity to churn based on their last five minutes of website activity."

2. Salesforce Help, Data Cloud, "Calculated Insights". The documentation clarifies, "Calculated insights run on a schedule that you define (hourly or daily)..." This contrasts with the streaming nature of the question's requirement.