Cecília Candolo

Poly-bagging predictors for classification modelling for credit scoring

Credit scoring modelling comprises one of the leading formal tools for supporting the granting of credit. Its core objective consists of the generation of a score by means of which potential clients can be listed in the order of the probability of default. A critical factor is whether a credit scoring model is accurate enough in order to provide correct classification of the client as a good or bad payer. In this context the concept of bootstraping aggregating (bagging) arises. The basic idea is to generate multiple classifiers by obtaining the predicted values from the fitted models to several replicated datasets and then combining them into a single predictive classification in order to improve the classification accuracy. In this paper we propose a new bagging-type variant procedure, which we call poly-bagging, consisting of combining predictors over a succession of resamplings. The study is derived by credit scoring modelling. The proposed poly-bagging procedure was applied to some different artificial datasets and to a real granting of credit dataset up to three successions of resamplings. We observed better classification accuracy for the two-bagged and the three-bagged models for all considered setups. These results lead to a strong indication that the poly-bagging approach may promote improvement on the modelling performance measures, while keeping a flexible and straightforward bagging-type structure easy to implement.

Electromyographic and magnetic resonance imaging evaluations of individuals with patellofemoral pain syndrome

OBJECTIVES To analyze the electrical activity of the vastus medialis obliquus (VMO), vastus lateralis longus (VLL) and vastus lateralis obliquus (VLO) muscles of individuals with patellofemoral pain syndrome (PFPS) during maximum voluntary isometric contraction (MVIC) of lower leg extension with the knee at 30 degrees; to assess pain using a visual analogue scale (VAS); and to assess patellar positioning using magnetic resonance imaging (MRI). METHODS Twelve women with PFPS and 12 clinically normal women were evaluated. They performed five MVICs of lower leg extension at 30 degrees for electromyographic (EMG) analysis. Using MRI, the sulcus angle (SA), congruence angle (CA), patellar tilt angle (PTA) and patellar displacement (PD) were obtained. The following statistical tests were used: analysis of variance (ANOVA) for repeated measurements to assess EMGs; Mann-Whitney U test to analyze MRIs; Pearsons (r) correlation test between EMGs and MRIs; and one-way ANOVA to evaluate pain (p < or = 0.05). RESULTS In the PFPS group, there was greater electrical activity in the VLL than in the VMO. In both groups, there was greater electrical activity in the VMO and VLL than in the VLO. In the PFPS group, the MRI showed higher SA and lower CA values, and there was a negative correlation between the VMO and the PTA. CONCLUSION The data suggest that, in individuals with PFPS, greater electrical activity in the VLL combined with an increased SA and a decreased CA may contribute to patellar instability.

Validity and Reliability of the 6-Minute Step Test in Healthy Individuals: A Cross-sectional Study.

Objective:To determine the 6-minute step tests (6MST) reliability and validity and to establish reference performance values of this test. Design:Prospective observational cross-sectional study. Setting:Spirometry and Respiratory Physiotherapy Laboratory, Federal University of São Carlos (institutional). Participants:Ninety-one individuals [42 men and 49 women, mean age = 39 years (SD, 17 years)] without any diagnosed diseases and with normal exercise capacity [6-minute walk test (6MWT) >75% of the predicted normal]. Independent Variables:Participants underwent two 6MST on 1 day and two 6MWT on another day in randomized order. Furthermore, age, gender, height, weight, lower limbs length, abdominal circumference, percentage of body fat, and fat-free mass were obtained. Main Outcome Measures:Test–retest reliability was assessed by comparing the findings of the two 6MST using the intraclass correlation coefficient (ICC) and Bland–Altman plot. Validity was assessed by comparing outcomes of the 6MST to outcomes of 6MWT using the Pearson correlation coefficient. A multiple regression analysis was conducted using the stepwise method to develop an equation to predict reference values. Results:The performance (mean steps ± SD) in the first and second test was 149 ± 34 and 149 ± 36 steps, respectively, which was correlated to distance (in meters) in 6MWT (r = 0.72; P < 0.05). Six-minute step test performance was reliable (ICC = 0.9; 95% confidence interval: 0.85–0.93). The equation to predict reference values for the first 6MST was significant (P < 0.001 and R2 = 0.48): Performance(steps) = 174 to 1.05 × Age(years) to women and Performance(steps) = 209 to 1.05 × Age(years) to men. Conclusions:Six-minute step test is a reliable and valid test. Moreover, the number of steps may be predicted by demographic and anthropometric variables with moderate strength of prediction. Clinical Relevance:Six-minute step test is an exercise test that is easy to be conducted, more tolerable than a graded exercise test, requires fewer equipments and space, and permits better monitoring of the participants. The assessment of the reliability, validity, and reference values will provide a better interpretability for clinicians to use it, especially in primary care.

Nível de atividade física e sua relação com quedas acidentais e fatores psicossociais em idosos de centro de convivência

This cross-sectional study aimed to estimate the level of physical activity (LPA) in Senior Citizen Centers (SCC) users and the possible association with accidental falls and other related variables. The sample included 291 senior citizens from three different SCC from Cuiaba City (Mato Grosso State, Brazil) who answered the Mini-mental State Examination-MMSE and the International Physical Activity Questionnaire-IPAQ. Descriptive Statistic and Ordinal Logistic Regression by Proportional Odds Model (POM) were used for data analyses (p<0.05). About 40% of the elderly suffered falls in the last year. Analyzing the LPA, 38.1% of voluntaries were classified as high level, 49.8% as moderate level and 12.1% as low level. The results of POM indicated that the oldest group showed lower chance to be classified as high LPA (OR=0.78: IC=0.96-2.92) than younger groups. These results were observed in seniors that do not have a job (OR=0.38: IC=0.22-0.67) and those with low cognitive status (OR=0.31; IC=0.31-0.38). Accidental falls occurrence did not show association with LPA.

Opportunistic Behavior and Stability of Governances in Automotive Fuel Negotiations in the State of São Paulo (Brazil)

This article evaluates the factors that influence the opportunistic behavior observed in the automotive fuel distribution sector (ethanol and gasoline) and the way this behavior has affected the stability of transactions between distributors and retail gas stations in the State of São Paulo (Brazil). The methodology we adopt in this study is logistic regression. Contracts analysis showed good results in reducing opportunism through fuel adulteration (-68.6 percent), and enforcement by distributors proved effective in curbing opportunism through noncompliance with agreements. We also found that time in the market is a factor conducive to opportunistic behavior, while the application of higher prices is a factor that discourages opportunism.

Exacerbation and functional capacity of patients with COPD undergoing an exercise training program: longitudinal study

Objective To analyze if there is influence of body weight, body mass index (BMI), body composition, dyspnoea, grip strength and tolerance to exertion in the occurrence of exacerbation during a 12-month follow up of patients with COPD who underwent a physical training program. Material and methods Sixty three patients were distributed in two groups, (Exacerbation Group — EG, n = 29; Non-Exacerbated Group — NEG, n = 34). The Mann Whitney test was used for the comparison between groups, the Friedman test (posthoc Dunn) to compare the assessments and the logistic regression analysis, with a significance level of p < 0.05. Results There is a significant difference between the groups in age and walked distance (WD) in the sixminute walk test (6MWT). The WD was reduced in 6th, 9th and 12th month revaluation compared to baseline and 3 months for the EG. Logistic regression analysis showed a significant interaction between the lean body mass and the WD, BMI with the lean body mass and the BMI with the WD, this and the isolated dyspnoea, and lean body mass with body weight. Conclusion Involving several variables along the follow up of patients with COPD in physical therapy programs is important, since it may prevent or reduce the chance of the occurrence of exacerbations. In addition, older patients with less tolerance to physical activity had a higher number of episodes of exacerbation, even when participating in a physiotherapy program associated to exercise training.