A novel technique for detecting sudden concept drift in healthcare data using multi-linear artificial intelligence...

Area of Science:

• Financial markets
• Data stream analysis
• Machine learning

Background:

• Financial markets generate massive daily data streams.
• Analyzing dynamic behaviors in these systems is challenging.
• Stabilizing financial systems is crucial for societal well-being.

Purpose of the Study:

• To propose a method for handling concept drift in financial data streams.
• To stabilize financial market system behavior through drift estimation.
• To improve the accuracy of concept drift detection.

Main Methods:

• Utilizing a sliding window approach for data stream processing.
• Applying the random forest algorithm for concept drift detection.
• Comparing the proposed method with existing techniques like OCDD, ADWIN, and Page-Hinckley test.

Main Results:

• The proposed approach demonstrates promising accuracy in detecting concept drift.
• It offers a robust solution for managing dynamic changes in financial data.
• Achieved superior performance compared to established drift detection methods.

Conclusions:

• The sliding window and random forest method effectively addresses concept drift in financial markets.
• This approach contributes to stabilizing financial systems and enhancing data analysis.
• The findings suggest a significant advancement in financial data stream analysis techniques.