Uma análise da ferramenta Ludwig, aplicada no contexto da ciência dos dados e modelos de deep learning

Abstract

Data science is directly associated to the context of increasing income of data, given that this subject aims to the understanding of it. Through this issue, information is extracted by several means in order to accomplish different interests. Considering the knowledge’s growing democratization process, some tools were developed with the purpose to help users on their data science projects. In this scenario, Uber elaborated Ludwig, a deep learning tool which doesn’t demand profound knowledge in programming skills. The aim of this study is to analyze it in the interest of measuring its potential by classification and regression models. The methods consist of installing Ludwig and using different approaches to verify its performance. In conclusion, although Ludwig fulfills its purpose on training and testing deep learning models without the need to write code, it still needs adjustments in order to become more attractive to unexperienced users.