| Popis: |
The aim of our study is to present an application of Generalized Linear Models to the prevention of the risk of insolvency of an automotive financial service. In order to forecast the payments (by installment) of the customers, we use a logit multivariate regression model. Before fitting the model, we resort to sample logits, Generalized Additive Models and univariate logistic regression in order to identify the subset of best predictors and verify the assumptions underpinning the statistical model. For the estimated model, we use the Wald statistics to assess the significance of the coefficients, the Likelihood Ratio to test the goodness of fit of the model, and the Odds Ratio to interpret the meaning of the estimated coefficients. In order to verify the goodness of fit of the model, we utilize classifications tables and the Receiver Operating Characteristic curve. Finally, we externally validate the fitted model by means of a predictive-test on a training set. |
