PROFESSIONAL EVENT KEYNOTE-SPEAKER
Concept Drifting in Big Data
We live in a dynamic world, where changes are a part of everyday life. When there is a shift in data, the classification or prediction models need to be adaptive to the changes. In data mining the phenomenon of change in data distribution over time is known as concept drift.
For example, click-streams of user's navigating news website may reflect the preferences of reading through the analysis systems. When people's interest of reading change, however, the old user's behaviour model is not applicable any more than the drifting of concepts appears. Traditional approaches don’t take into consideration the fact that user’s interests keep on changing with respect to time and latest reading behaviour, which reflects the user’s current interest, should be given more importance over old behaviour.
The same applies to many other non-stationary data like stock marketing, weather and customer preferences.
To know more about this visit our Conference AI 2020