In an extensive simulation study, we show how the increased predictive accuracy of the Grabit model compared to other classifiers depends on (i) the correlation between the auxiliary data and the latent decision function, (ii) the class imbalance ratio, (iii) the sample size, and (iv) the complexity of the decision function. In our application, we show that the Grabit model considerably outperforms other state-of-the-art approaches for default prediction of loans made to Swiss small and medium-sized enterprises (SMEs). The problem of class imbalance is aggravated if the size of the data set is small. In general, class imbalance, i.e., the situation where a majority class is much more frequent than a minority class, is a problem not just in the area of default prediction but also for various other classification tasks such as fraud detection or predictive maintenance. A common problem is that default events are relatively rare and the number of defaulted companies is typically much lower than the number of non-defaults. Statistical default prediction is traditionally approached using binary classification models see, e.g., Bauer and Agarwal and Tian et al. Statistical modeling of default began in the late 1960s with the seminal papers of Beaver and Altman. As a result, default prediction has been of major interest to both researchers and practitioners in the financial sector for almost a century. Prediction of corporate failures is important since bankruptcies can result in significant economic losses for investors and even cause economic downturns and recessions. We obtain a large improvement in predictive performance compared to other We apply the Grabit model for predictingĭefaults on loans made to Swiss small and medium-sized enterprises (SME), and Transformations for the covariates, and its predictive performance is not It is robust against outliers in covariates and scale invariant to monotonic Grabit model can account for general forms of non-linearities and interactions, Examples of this include censored data, fractional responseĭata, corner solution response data, rainfall data, and binary classificationĭata where additional information, that is related to the underlyingĬlassification mechanism, is available. The so called Grabit model allows for modelingĭata that consist of a mixture of a continuous part and discrete point massesĪt the borders. G.Harris -Preceding unsigned comment added by 78.16.151.We introduce a novel model which is obtained by applying gradient treeīoosting to the Tobit model. No response to emails? Not true ! Sure a few days delay but Ilan personally responds in a very friendly and helpful manner In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.īetacommandBot ( talk) 23:36, 13 February 2008 (UTC) Reply I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. Image:Grabitmain.png is being used on this article. Thank you.īetacommandBot ( talk) 23:35, 13 February 2008 (UTC) Reply Fair use rationale for Image:Grabitmain.png If you have any questions please ask them at the Media copyright questions page.

Note that any fair use images lacking such an explanation can be deleted one week after being tagged, as described on criteria for speedy deletion.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Do not simply insert a blank template on an image page. Using one of the templates at Wikipedia:Fair use rationale guideline is an easy way to ensure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Please go to the image description page and edit it to include a fair use rationale. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use. Image:Grabitlogo.png is being used on this article.

Grabit media