Citas bibligráficas
Rafael, O., (2023). Regresión a la media para datos fraccionarios usando enlaces asimétricos con una aplicación a cobranzas de tarjetas de crédito [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/27259
Rafael, O., Regresión a la media para datos fraccionarios usando enlaces asimétricos con una aplicación a cobranzas de tarjetas de crédito [Tesis]. PE: Universidad Nacional de Ingeniería; 2023. http://hdl.handle.net/20.500.14076/27259
@misc{renati/713014,
title = "Regresión a la media para datos fraccionarios usando enlaces asimétricos con una aplicación a cobranzas de tarjetas de crédito",
author = "Rafael Taipe, Osman Dennys",
publisher = "Universidad Nacional de Ingeniería",
year = "2023"
}
In this research a new regression model to the mean is proposed, for fractional independent variables using asymmetric links, these variables take continuous valúes in the open interval (0,1) and even taking the extreme values 0 and 1, these variables could be proportions, rates, ratios, etc. For the development of the thesis, we take as a basis the research developed by (Bayes & Valdivieso, 2016) that adjusts the set of equations for the range (0,1) and the extremes 0 and 1 in a single process of estimating the mean g(y) = x'P, so it can be worked with arbitrary link functions. For this special case (Bayes & Valdivieso, 2016) for convenience they used a logistic type link, where g is a link function and x is a vector of covariates. The proposed links are of type Loglog and Cloglog derived from the Gumbell Standard Complementary Exponentiated distribution, which was worked on by (Gupta & Kundu, 1999) under the assumption of a better fit to the data. The optimization of the log-likelihood function was developed using a Fisher scoring algorithm modifying using a matlab routine with the analytical gradient vector and the Hessian matrix. To test the hypotheses, a comparative simulation analysis was developed, with the objective of recovering the parameters and comparing the ECM for the comparison of the proposed model vs the model of (Bayes & Valdivieso, 2016). Additionally, the developed model was applied to data on credit card collections for a Peruvian bank.
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons