Credit Scoring Using Binary Quantile Regression

Kordas, Gregory

doi:10.1007/978-3-0348-8201-9_10

Cited by 9 publications

(6 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In real applications, y is the observed binary response and the interest is to predict y from knowledge of x. y ⋆ is unobserved and used mainly for modelling purposes. Some examples of y ⋆ include the actual birth weight of babies in a study where the aim is to investigate the factors behind the birth of premature babies, the credit risk of a customer in a study where the aim is to discriminate between good and bad customers (Kordas 2002) or the willingness to participate to work in a study where the factors behind the decision to work or not are investigated (Kordas 2006).…”

Section: Binary Quantile Group Lassomentioning

confidence: 99%

Quantile regression with group lasso for classification

Hashem

Vinciotti

Alhamzawi

et al. 2015

Adv Data Anal Classif

View full text Add to dashboard Cite

Applications of regression models for binary response are very common and models specific to these problems are widely used. Quantile regression for binary response data has recently attracted attention and regularized quantile regression methods have been proposed for high dimensional problems. When the predictors have a natural group structure, such as in the case of categorical predictors converted into dummy variables, then a group lasso penalty is used in regularized methods. In this paper, we present a Bayesian Gibbs sampling procedure to estimate the parameters of a binary quantile regression model under a group lasso penalty. Simulated and real data show a good performance of the proposed method in comparison to mean-based approaches and to quantile-based approaches which do not exploit the group structure of the predictors.

show abstract

Section: Binary Quantile Group Lassomentioning

confidence: 99%

Quantile regression with group lasso for classification

Hashem

Vinciotti

Alhamzawi

et al. 2015

Adv Data Anal Classif

View full text Add to dashboard Cite

show abstract

“…The high dimensionality of the feature space in this problem prevents binary quantile regression [10,11] from being applicable. To test the Median Boost algorithm on the building block labeling task, we used the standard normal cumulative distribution function with h = 0.1 as the approximation to the indicator function.…”

Section: Resultsmentioning

confidence: 99%

“…Eqn. (10) indicates that maximizing Eqn. (5) is equivalent to maximizing the sum of T P and T N, i.e., the number of correctly classified examples.…”

Section: Median Classifiermentioning

confidence: 99%

“…To guarantee the performance of the resulting model, ν is fixed at a small value as suggested by [1,5]. The median classifier has close connection to binary quantile regression [10,11], which optimizes an objective function similar to Eqn. (5).…”

Section: Median Boost: Boosting Based Median Classifiermentioning

confidence: 99%

“…If the feature vector has long tail distribution, or there are outliers in the data, it is well-known that the median is usually more stable than other statistics [7,8,10,11]. In particular, if some of the components of the feature vector have high variability then, regardless of whether or not these components convey information about the class label, classification accuracy might be poor.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Labeling Image Patches by Boosting based Median Classifier

Zheng¹

2011

Procedings of the British Machine Vision Conference 2011

View full text Add to dashboard Cite

This paper presents a median based classifier which predicts the conditional median of the class label given the feature vector. The class label is defined through a hidden variable whose median is further modeled as an additive model of the feature vector. We propose to estimate the conditional median of the hidden variable given the feature vector in the framework of generic functional gradient algorithm [5]. An equivalent form of the definition of median is introduced, whose smoothed version is employed as the objective function. To fit the model for the conditional median, the proposed objective function is maximized by gradient ascent in functional space, updating the fitted model a small step in the gradient direction in each iteration. The resulting algorithm, Median Boost, is a boosting like procedure which obtains the informative features and the classifier at the same time. On the task of labeling building blocks in natural images, the comparison results show that Median Boost performs better than or similar to several alternatives.

show abstract

Additional models

2014

Quantile Regression

View full text Add to dashboard Cite

Credit Scoring Using Binary Quantile Regression

Cited by 9 publications

References 4 publications

Quantile regression with group lasso for classification

Quantile regression with group lasso for classification

Labeling Image Patches by Boosting based Median Classifier

Additional models

Contact Info

Product

Resources

About