On the impact of disproportional samples in credit scoring models: An application to a Brazilian bank data

Statistical methods have been widely employed to assess the capabilities of credit scoring classification models in order to reduce the risk of wrong decisions when granting credit facilities to clients. The predictive quality of a classification model can be evaluated based on measures such as sensitivity, specificity, predictive values, accuracy, correlation coefficients and information theoretical measures, such as relative entropy and mutual information. In this paper we analyze the performance of a naive logistic regression model (Hosmer & Lemeshow, 1989) and a logistic regression with state-dependent sample selection model (Cramer, 2004) applied to simulated data. Also, as a case study, the methodology is illustrated on a data set extracted from a Brazilian bank portfolio. Our simulation results so far revealed that there is no statistically significant difference in terms of predictive capacity between the naive logistic regression models and the logistic regression with state-dependent sample selection models. However, there is strong difference between the distributions of the estimated default probabilities from these two statistical modeling techniques, with the naive logistic regression models always underestimating such probabilities, particularly in the presence of balanced samples. (C) 2012 Elsevier Ltd. All rights reserved.

Genetic relatedness of Plasmodium falciparum isolates and the origin of allelic diversity at the merozoite surface protein-1 (MSP-1) locus in Brazil and Vietnam

Abstract Background Despite the extensive polymorphism at the merozoite surface protein-1 (MSP-1) locus of Plasmodium falciparum, that encodes a major repetitive malaria vaccine candidate antigen, identical and nearly identical alleles frequently occur in sympatric parasites. Here we used microsatellite haplotyping to estimate the genetic distance between isolates carrying identical and nearly identical MSP-1 alleles. Methods We analyzed 28 isolates from hypoendemic areas in north-western Brazil, collected between 1985 and 1998, and 23 isolates obtained in mesoendemic southern Vietnam in 1996. MSP-1 alleles were characterized by combining PCR typing with allele-specific primers and partial DNA sequencing. The following single-copy microsatellite markers were typed : Polyα, TA42 (only for Brazilian samples), TA81, TA1, TA87, TA109 (only for Brazilian samples), 2490, ARAII, PfG377, PfPK2, and TA60. Results The low pair-wise average genetic distance between microsatellite haplotypes of isolates sharing identical MSP-1 alleles indicates that epidemic propagation of discrete parasite clones originated most identical MSP-1 alleles in parasite populations from Brazil and Vietnam. At least one epidemic clone propagating in Brazil remained relatively unchanged over more than one decade. Moreover, we found no evidence that rearrangements of MSP-1 repeats, putatively created by mitotic recombination events, generated new alleles within clonal lineages of parasites in either country. Conclusion Identical MSP-1 alleles originated from co-ancestry in both populations, whereas nearly identical MSP-1 alleles have probably appeared independently in unrelated parasite lineages.

Targeting the non-structural protein 1 from dengue virus to a dendritic cell population confers protective immunity to lethal virus challenge

Dengue is the most prevalent arboviral infection, affecting millions of people every year. Attempts to control such infection are being made, and the development of a vaccine is a World Health Organization priority. Among the proteins being tested as vaccine candidates in preclinical settings is the non-structural protein 1 (NS1). In the present study, we tested the immune responses generated by targeting the NS1 protein to two different dendritic cell populations. Dendritic cells (DCs) are important antigen presenting cells, and targeting proteins to maturing DCs has proved to be an efficient means of immunization. Antigen targeting is accomplished by the use of a monoclonal antibody (mAb) directed against a DC cell surface receptor fused to the protein of interest. We used two mAbs (αDEC205 and αDCIR2) to target two distinct DC populations, expressing either DEC205 or DCIR2 endocytic receptors, respectively, in mice. The fusion mAbs were successfully produced, bound to their respective receptors, and were used to immunize BALB/c mice in the presence of polyriboinosinic: polyribocytidylic acid (poly (I:C)), as a DC maturation stimulus. We observed induction of strong anti-NS1 antibody responses and similar antigen binding affinity irrespectively of the DC population targeted. Nevertheless, the IgG1/IgG2a ratios were different between mouse groups immunized with αDEC-NS1 and αDCIR2-NS1 mAbs. When we tested the induction of cellular immune responses, the number of IFN-γ producing cells was higher in αDEC-NS1 immunized animals. In addition, mice immunized with the αDEC-NS1 mAb were significantly protected from a lethal intracranial challenge with the DENV2 NGC strain when compared to mice immunized with αDCIR2-NS1 mAb. Protection was partially mediated by CD4(+) and CD8(+) T cells as depletion of these populations reduced both survival and morbidity signs. We conclude that targeting the NS1 protein to the DEC205(+) DC population with poly (I:C) opens perspectives for dengue vaccine development.