Risk assessment of credit portfolios is of pivotal importance in the banking industry. The bank that has the most accurate view of its credit risk will be the most profitable. One of the main pillars in the assessing credit risk is the estimated probability of default of each counterparty, ie the probability that a counterparty cannot meet its payment obligations in the horizon of one year. A credit rating system takes several characteristics of a counterparty as inputs and assigns this counterparty to a rating class. In essence, this system is a classifier whose classes lie on an ordinal scale.
In this paper we apply linear regression ordinal logistic regression, and support vector machine techniques to the credit rating problem. The latter technique is a relatively new machine learning technique that was originally designed for the two-class problem. We propose two new techniques that incorporate the ordinal character of the credit rating problem into support vector machines. The results of our newly introduced techniques are promising.
In the study presented, different hybrid model approaches are proposed for reservoir inflow modeling from the meteorological data (monthly precipitation, one-month-ahead precipitation and monthly mean temperature data) by the combined use of discrete wavelet transform (DWT) and different black box techniques. Multiple linear regression (MLR), feed forward neural networks (FFNN) and least square support vector machines (LSSVM) were considered as the black box methods. In the modeling strategy, meteorological input data were decomposed into wavelet sub-time series at three resolution levels and ineffective sub-time series were eliminated by Mallows’ Cp based all possible regression method. As a result of all possible regression analyses, 2-months mode of time series of monthly temperature (D1_Tt), 8-months mode of time series (D3_Tt) of monthly temperature and approximation mode of time series (A3_Tt) of monthly temperature were eliminated. Remained effective sub-time series were used as the inputs of MLR, FFNN and LSSVM. When the performances of the training and testing periods were compared, it was observed that the DWTFFNN conjunction model has better results in terms of mean square errors (MSE) and determination coefficients (R2 ) statistics. The discrete wavelet transform approach also increased the accuracy of multiple linear regression and least squares support vector machines.