Banca de QUALIFICAÇÃO: LAION LIMA BOAVENTURA

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : LAION LIMA BOAVENTURA
DATE: 01/12/2023
TIME: 14:00
LOCAL: Sala 406 do Bloco B do Campus de Santo André da Universidade Federal do ABC
TITLE:

Clustered Spearman-Markowitz Method: A New Hybrid Approach for Resource Selection in Credit Granting Models

PAGES: 51
BIG AREA: Outra
AREA: Multidisciplinar
SUMMARY:

Democratization and free access to large volumes of data have made studies on feature selection methods a catalyst in directing efforts to exclude redundant data from the process of building a machine learning model, in order to make it is less likely to suffer from overfitting, as well as simpler and, consequently, less costly to monitor and consume. In the literature, there are basically four approaches to resource selection methods: embedded, filter, wrapper and the hybrid method, the result of the combination of two of the first three. This work proposes a new hybrid approach, aimed at solving problems involving predictive credit scoring models, which combines concepts from the Markowtiz Model for portfolio allocation to the Spearman Correlation Coefficient, so that the resulting method is competitive with traditionally used methods. The validation and comparison process of the proposed method, called the Clustered Spearman-Markowitz Method (SMC), was carried out through Monte Carlo simulation studies on synthetic credit scoring data, as well as through an application to a set of real data from a Credit Bureau.

COMMITTEE MEMBERS:
Presidente - Interno ao Programa - 2090028 - FILIPE IEDA FAZANARO
Membro Titular - Examinador(a) Interno ao Programa - 2334927 - ANDRE KAZUO TAKAHATA
Membro Titular - Examinador(a) Externo ao Programa - 3044516 - LUNEQUE DEL RIO DE SOUZA E SILVA JUNIOR
Membro Suplente - Examinador(a) Interno ao Programa - 1761107 - RICARDO SUYAMA