Home › Key concepts › Data leakage

Key concept · Crawl (beginner) · the same on every platform

Data leakage

When the model is secretly given the answer during training.

What it is

Leakage happens when a feature contains information that would not be known at prediction time, such as a 'cancellation date' column when predicting whether someone will cancel.

Why it matters

Leaky models score brilliantly in testing and fail completely in real use.

An example

A churn model scores 99% because 'days since last login' was computed after the customer had already left.

The common mistake

Celebrating a too-good score instead of hunting for the leak.

Where you meet it on each platform

The idea stays the same; only the product name changes.

JobAWSAzureGoogle CloudDatabricksOpen source
Feature storethe same feature values for training and for live predictionsSageMaker Feature StoreAzure ML feature storeVertex AI Feature StoreFeature engineering in Unity CatalogFeast

Related ideas

Where it shows up

Part of the Crawl level on these paths: ML Engineering (learning basics).