Data Drift

A type of model drift that occurs when changing business conditions diminish the accuracy of machine learning models over time.

Added Perspectives
Most importantly, data teams must rinse and repeat. They must identify data drift—i.e., changes in market conditions or other aspects of your environment—then pull their ML models out of production, re-train those models and re-implement them. Figure 1 illustrates the three stages of the ML lifecycle.
(Blog)
After consulting dozens of enterprises in stealth mode, Pancha and Prabhakar launched the StreamSets Data Collector product in 2015 to automate the data ingestion process for data engineers. They also sought to automatically detect and manage “data drift”—those gradual changes to source schema, metadata, and semantics that typically force teams to manually reconfigure data pipelines.
- Kevin Petrie in Deep Dive: Hybrid Cloud Data Integration May 18, 2021
(Report)
  • 0..9
  • A
  • B
  • C
  • D
  • E
  • F
  • G
  • H
  • I
  • J
  • K
  • L
  • M
  • N
  • O
  • P
  • Q
  • R
  • S
  • T
  • U
  • V
  • W
  • X
  • Y
  • Z