Prediction of personal default risks based on a sparrow search algorithm with support vector machine model.

Area of Science:

• Machine Learning
• Financial Risk Management
• Computational Intelligence

Background:

• Accurate personal credit evaluation is crucial for commercial banks to mitigate financial losses.
• Traditional credit scoring models often struggle with complex, high-dimensional datasets.
• Machine learning offers promising avenues for enhancing default risk prediction accuracy.

Purpose of the Study:

• To develop and evaluate a novel hybrid model for predicting personal credit default risk.
• To investigate the efficacy of combining the Sparrow Search Algorithm (SSA) with Support Vector Machines (SVM) for this task.
• To demonstrate the practical value of the proposed SSA-SVM model in commercial banking.

Main Methods:

• Utilized personal credit data for analysis.
• Applied statistical analysis, data normalization, and principal factor analysis for data preprocessing.
• Developed and implemented a hybrid SSA-SVM model for default risk prediction.
• Compared the performance of the SSA-SVM model against the standard SVM model.

Main Results:

• Data preprocessing techniques significantly improved evaluation index performance compared to raw data.
• The SSA-SVM model consistently outperformed the standard SVM model across various evaluation metrics.
• The hybrid model demonstrated enhanced accuracy in predicting personal default risk.

Conclusions:

• The proposed data processing methods are effective for personal credit data.
• The SSA-SVM hybrid model offers a significant improvement over the SVM model for personal default risk prediction.
• The SSA-SVM model holds considerable practical value for commercial banks seeking to refine their credit evaluation processes.