Loading...
Publication
MLCASE: Plataforma Automatizada de Desenvolvimento de Modelos de Machine Learning
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 8.43 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O desenvolvimento de modelos de Machine Learning é um processo intrinsecamente
complexo, composto por múltiplas etapas, como o pré-processamento de dados, a seleção de
algoritmos, a otimização de hiperparâmetros e a avaliação de desempenho. Este projeto
propõe o desenvolvimento de uma plataforma automatizada, denominada MLCASE, com o
objetivo de simplificar e acelerar estas etapas, permitindo que os cientistas de dados
otimizem o seu tempo e concentrem os seus esforços na análise de resultados e na obtenção
de insights estratégicos. A MLCASE foi concebida com base na framework Streamlit,
proporcionando uma interface intuitiva e interativa, enquanto automatiza tarefas
fundamentais como a seleção de features, a pesquisa e otimização de hiperparâmetros e a
avaliação de modelos de aprendizagem automática. Adicionalmente, a plataforma integra a
criação automática de relatórios técnicos e gráficos analíticos, promovendo a
democratização do acesso a técnicas avançadas de Machine Learning e a sua aplicação
eficiente em áreas como saúde, finanças, marketing e ciência social.
The development of Machine Learning models is inherently complex, encompassing multiple stages such as data preprocessing, algorithm selection, hyperparameter optimization, and performance evaluation. This project proposes the development of an automated platform named MLCASE, designed to simplify and accelerate these stages, enabling data scientists to optimize their time and focus on result analysis and strategic insights. MLCASE is built using the Streamlit framework, providing an intuitive and interactive interface while automating critical tasks such as feature selection, hyperparameter tuning, and machine learning model evaluation. Additionally, the platform includes automated creation of technical reports and analytical visualizations, promoting the democratization of access to advanced Machine Learning techniques and their efficient application in fields such as healthcare, finance, marketing, and social sciences.
The development of Machine Learning models is inherently complex, encompassing multiple stages such as data preprocessing, algorithm selection, hyperparameter optimization, and performance evaluation. This project proposes the development of an automated platform named MLCASE, designed to simplify and accelerate these stages, enabling data scientists to optimize their time and focus on result analysis and strategic insights. MLCASE is built using the Streamlit framework, providing an intuitive and interactive interface while automating critical tasks such as feature selection, hyperparameter tuning, and machine learning model evaluation. Additionally, the platform includes automated creation of technical reports and analytical visualizations, promoting the democratization of access to advanced Machine Learning techniques and their efficient application in fields such as healthcare, finance, marketing, and social sciences.
Description
Keywords
Aprendizagem automática AutoML Desenvolvimento de modelos Automatização Otimização de hiperparâmetros Streamlit