← Library · Core concept
Concept Drift
Concept drift refers to the phenomenon where the statistical properties of the target variable, which the model is trying to predict, change over time in ways that were not anticipated by the model at the time of training. This means the relationship between the input features and the output changes. It can be gradual, abrupt, or recurring, directly impacting a model's performance in real-world scenarios.
In plain terms
It's like trying to navigate with an old map after entire cities and roads have changed, making your directions increasingly unreliable.
Why it matters
Ignoring concept drift leads to stale and ineffective AI models, necessitating continuous monitoring and retraining strategies.
Learn one new AI thing every day.
Daily Deck sends you seven plain-English cards like this every morning. Free.
Start free