Semnan University Press Journal of Modeling in Engineering 2008-4854 23 Special Issue 81 2025 06 22 IDCOST: A Method for Increasing Data Criterion Service by Scoring Credit Imbalanced Data Using Applied SVM IDCOST: A Method for Increasing Data Criterion Service by Scoring Credit Imbalanced Data Using Applied SVM 1 18 9519 10.22075/jme.2025.31252.2493 FA Arash Ghorbannia Delavar Department of Computer Engineering and Information Technology, Payam Noor University, Tehran, Iran Sadaf Sadat Ziya Department of Computer Engineering and Information Technology, Payam Noor University, Tehran, Iran Journal Article 2023 07 13 Unbalanced credit data can pose significant challenges in applied data mining. To address this, we propose a method that utilizes a scoring technique and support vector machine (SVM) to enhance data criterion service. Our approach integrates index feature selection and IDCOST method, which reduces data redundancy and balances feature selection data sets with a valid index. We also use feature selection and kernel modification to improve accuracy while reducing computational complexity and execution time. Our proposed method can detect credit card fraud and credit card default data sets with higher sensitivity than other methods. It presents a promising solution for tackling credit data issues in applied SVM data mining and has the potential to improve data analysis accuracy and reduce computational complexity in various fields.<br />The IDCOST method is presented in pre-processing, training, validation, and testing stages. We use detector threshold clustering in the pre-processing stage, sensitivity and feature validation on the models in the training stage, and score each sample in the test dataset in the testing stage. The proposed method's accuracy is optimized by selecting an appropriate cluster head in data classification and employing a scoring technique. In conclusion, our proposed method is an effective solution for tackling credit data issues in applied SVM data mining. By integrating index feature selection, IDCOST method, feature selection, and kernel modification, we can accurately detect credit card fraud and credit card default data sets while reducing data redundancy and computational complexity. Unbalanced credit data can pose significant challenges in applied data mining. To address this, we propose a method that utilizes a scoring technique and support vector machine (SVM) to enhance data criterion service. Our approach integrates index feature selection and IDCOST method, which reduces data redundancy and balances feature selection data sets with a valid index. We also use feature selection and kernel modification to improve accuracy while reducing computational complexity and execution time. Our proposed method can detect credit card fraud and credit card default data sets with higher sensitivity than other methods. It presents a promising solution for tackling credit data issues in applied SVM data mining and has the potential to improve data analysis accuracy and reduce computational complexity in various fields.<br />The IDCOST method is presented in pre-processing, training, validation, and testing stages. We use detector threshold clustering in the pre-processing stage, sensitivity and feature validation on the models in the training stage, and score each sample in the test dataset in the testing stage. The proposed method's accuracy is optimized by selecting an appropriate cluster head in data classification and employing a scoring technique. In conclusion, our proposed method is an effective solution for tackling credit data issues in applied SVM data mining. By integrating index feature selection, IDCOST method, feature selection, and kernel modification, we can accurately detect credit card fraud and credit card default data sets while reducing data redundancy and computational complexity. https://modelling.semnan.ac.ir/article_9519_d90b0af883e4d8143534a42089fc01bb.pdf