Data drift is a term used in the fields of artificial intelligence, big data, smart data and digital transformation. It describes a change in the data that is used for an automated system or artificial intelligence.
Imagine a company uses an AI to predict the demand for products. The AI has trained the system on past data, e.g. customers buy more umbrellas at certain times of the year. But suddenly customers' purchasing habits change, for example because a longer summer brings less rain. The old data no longer matches the current conditions.
This is data drift: The data with which the system was originally trained differs from the current data over time. As a result, the AI's predictions or decisions become unreliable or even incorrect.
It is important to recognise data drift, as otherwise the companies concerned will be acting on the basis of outdated information. Monitoring and regularly adapting algorithms ensures that artificial intelligence and big data solutions work correctly even under changing conditions.
Deutsch 
English (UK)