Multiclass Bilateral-Weighted Fuzzy Support Vector Machine to Evaluate Financial Strength Credit Rating

Due to recent financial cries, changing global economic, business, geographic conditions, and regulatory concerns, financial intermediaries´ credit rating assessment is an area of soaring interest in both the academic world and the business community. In this work, we propose a new fuzzy support vector machine (FSVM) to discriminate different financial strength credit rating classifications of financial institutions (FIs) from one and another using one against-all FSVM approach. In this work we have developed the FSVM and consider the possibility that any FI rated in one class may also contain partial fulfillment for some alternate rating. Therefore, this new fuzzy technique for multiclass SVM is termed as multiclass bilateral-weighted fuzzy support vector machine (MCB-FSVM). For simplicity, we have considered only five basic credit ratings from best (A) to worst (E). Moreover, fuzzy class memberships are assigned for each FI based on credit ratings of previous year.