Design and analysis of management platform based on financial big data

Area of Science:

• Computer Science
• Data Science
• Financial Technology

Background:

• Traditional financial accounting faces limitations with new technologies and market development.
• Current financial data classification systems exhibit high error rates, hindering business value generation.
• Improving financial big data utilization is crucial for enhancing financial management information application.

Purpose of the Study:

• To develop an automated financial data classification system using fuzzy clustering.
• To implement abnormal data detection using the neighborhood relation-based local outlier factor (NLOF) algorithm.
• To design a financial data management platform integrating these algorithms on a distributed Hadoop architecture.

Main Methods:

• Fuzzy clustering algorithm for automated financial data classification.
• Local Outlier Factor (LOF) algorithm with neighborhood relation (NLOF) for anomaly detection.
• Distributed Hadoop architecture with MapReduce framework for parallel processing of algorithms.

Main Results:

• The proposed platform demonstrates improved algorithm performance and accuracy in financial data classification.
• Parallel execution of fuzzy clustering and LOF algorithms via MapReduce enhances processing efficiency.
• Experimental results show superior running efficiency and classification accuracy compared to existing methods.

Conclusions:

• The developed financial data management platform effectively addresses the limitations of traditional accounting.
• The integration of fuzzy clustering and LOF algorithms on Hadoop significantly improves financial data processing.
• The platform offers a superior and effective solution for enterprise financial data management and analysis.