Credit Scoring using Aggregation: An Empirical Study
When it comes to the area of finance, Credit Scoring has been regarded as one of the most important appraisal
tools of institutions in the last few decades. A number of statistical models are being used for credit scoring using a lot many
prediction techniques. In this paper, we propose an ensemble technique that aggregates a number of existing models such as
Random Forests, Support Vector Machine (SVM), Logistic Regression and Artificial Neural Nets, in order to better predict
credit scores and obtain a much higher accuracy rate than these individual techniques. A comparative analysis of various
traditional models, as well as the aggregated model is also provided.
Keywords - Credit scoring, Random Forests, SVM, Logistic Regression, Neural networks, Bagging