Risco de crédito: uma abordagem utilizando análise discriminante, regressão logística e redes neurais artificiais

Considering the relevance of researches concerning credit risk, model diversity and the existent indicators, this thesis aimed at verifying if the Fleuriet Model contributes in discriminating Brazilian open capital companies in the analysis of credit concession. We specifically intended to i) identify the economic-financial indicators used in credit risk models; ii) identify which economic-financial indicators best discriminate companies in the analysis of credit concession; iii) assess which techniques used (discriminant analysis, logistic regression and neural networks) present the best accuracy to predict company bankruptcy. To do this, the theoretical background approached the concepts of financial analysis, which introduced themes relative to the company evaluation process; considerations on credit, risk and analysis; Fleuriet Model and its indicators, and, finally, presented the techniques for credit analysis based on discriminant analysis, logistic regression and artificial neural networks. Methodologically, the research was defined as quantitative, regarding its nature, and explanatory, regarding its type. It was developed using data derived from bibliographic and document analysis. The financial demonstrations were collected by means of the Economática ® and the BM$FBOVESPA website. The sample was comprised of 121 companies, being those 70 solvents and 51 insolvents from various sectors. In the analyses, we used 22 indicators of the Traditional Model and 13 of the Fleuriet Model, totalizing 35 indicators. The economic-financial indicators which were a part of, at least, one of the three final models were: X1 (Working Capital over Assets), X3 (NCG over Assets), X4 (NCG over Net Revenue), X8 (Type of Financial Structure), X9 (Net Thermometer), X16 (Net Equity divided by the total demandable), X17 (Asset Turnover), X20 (Net Equity Profitability), X25 (Net Margin), X28 (Debt Composition) and X31 (Net Equity over Asset). The final models presented setting values of: 90.9% (discriminant analysis); 90.9% (logistic regression) and 97.8% (neural networks). The modeling in neural networks presented higher accuracy, which was confirmed by the ROC curve. In conclusion, the indicators of the Fleuriet Model presented relevant results for the research of credit risk, especially if modeled by neural networks.

Regressão Simplex aplicada a delineamentos de mistura e utilização do Algoritmo Boosting

In the composition of this work are present two parts. The first part contains the theory used. The second part contains the two articles. The first article examines two models of the class of generalized linear models for analyzing a mixture experiment, which studied the effect of different diets consist of fat, carbohydrate, and fiber on tumor expression in mammary glands of female rats, given by the ratio mice that had tumor expression in a particular diet. Mixture experiments are characterized by having the effect of collinearity and smaller sample size. In this sense, assuming normality for the answer to be maximized or minimized may be inadequate. Given this fact, the main characteristics of logistic regression and simplex models are addressed. The models were compared by the criteria of selection of models AIC, BIC and ICOMP, simulated envelope charts for residuals of adjusted models, odds ratios graphics and their respective confidence intervals for each mixture component. It was concluded that first article that the simplex regression model showed better quality of fit and narrowest confidence intervals for odds ratio. The second article presents the model Boosted Simplex Regression, the boosting version of the simplex regression model, as an alternative to increase the precision of confidence intervals for the odds ratio for each mixture component. For this, we used the Monte Carlo method for the construction of confidence intervals. Moreover, it is presented in an innovative way the envelope simulated chart for residuals of the adjusted model via boosting algorithm. It was concluded that the Boosted Simplex Regression model was adjusted successfully and confidence intervals for the odds ratio were accurate and lightly more precise than the its maximum likelihood version.