liveBook · Manning

Aggregate feature

As explained in Machine Learning System Design

A feature computed by summarizing multiple observations over a specified set or time window.

An aggregate feature summarizes multiple observations over a defined set or time window. In time-series work, it can describe recent behavior using values such as a mean, maximum, or minimum. section 11.1.2

Common aggregate features include:

  • Mean, maximum, or minimum over a specific time window. section 11.1.2
  • Standard deviation or variance computed from past data across different time periods. section 11.1.2
  • A feature generated alongside lags and other transformations for time-series modeling. section 11.1.2

How Machine Learning System Design covers it

How deep it goes
Defines windowed summaries and names common statistics, but stops short of implementation procedures or leakage safeguards.
Examples
Uses mean, maximum, minimum, standard deviation, and variance calculated over specified past-data periods.
What it assumes
Requires observations organized over time and a chosen set or time window to summarize.
What you can do after
Build it from scratch by calculating summary statistics over defined time windows for model features.
How it connects
Connects aggregates with time-series data, lag features, statistical features, and feature engineering.
Cover of Machine Learning System Design
liveBookmanning.com
Read Machine Learning System Design on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Machine Learning System Design. The section links point to the parts of the book it draws on.