Data mining algorithm in the identification of accounting fraud by smart city information technology

Area of Science:

• Accounting
• Computer Science
• Data Mining

Background:

• Traditional accounting fraud detection methods lack objectivity and accuracy.
• Stakeholder interests and company development necessitate effective fraud identification.

Purpose of the Study:

• To enhance the accuracy, efficiency, and objectivity of accounting fraud identification.
• To research data mining algorithms for accounting fraud detection using smart city information technology.

Main Methods:

• Utilized k-means clustering algorithm for data mining-based fraud identification.
• Collected and analyzed financial data from 62 fraudulent and 84 non-fraudulent listed companies (2012-2021).
• Employed electronic data collection, analysis, and retrieval systems from stock exchange websites.

Main Results:

• The data mining approach demonstrated improved adaptability to various fraud types.
• Reduced the comprehensive misjudgment rate by 3% compared to traditional methods.
• ROC curve analysis indicated superior performance of the proposed method.

Conclusions:

• Data mining algorithms integrated with smart city information technology enhance accounting fraud detection accuracy.
• The proposed method improves audit objectivity and effectiveness.
• This approach can uncover novel fraud patterns missed by rule-based systems.