Comparativo de alguns modelos de machine learning utilizando dados de domínio público e a linguagem python

• Autor(a) principal: Viana, Wesley Muller Oliveira
• Data de Publicação: 2021
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://hdl.handle.net/11449/214516
• topic: Naive Bayes; Machine learning; Logistic regression; Naive-bayes; Decision tree; Transaction classification; Machine Learning; Regressão Logística; SVM; Árvore de decisão; KNN; Classificação de transações

There are many applications of machine learning models in various areas, and it is an area of research with continuous development. This work proposes a study of 5 classifier models with supervised learning. Using the Python programming language and some libraries allows the improvement in the approach to prevent credit card fraud. For the development of the work, a public database of European credit card transactions was used to minimize the data modeling problems arising from the imbalanced data. Two data split techniques were also tested (undersampling and oversampling) to obtain the best performance of the models analyzed. This performance was evaluated using some performance metrics such as accuracy, f1-score, precision, and recall. The academic work explores possible improvements for further work, such as dimensionality reduction of the feature dataset, parameter optimization of the tested models, and regularization of the modeling using penalization for the optimizers.