Have you ever wondered how much any data is actually valuable? This project aims to provide a useful tool for data analysis, with a simple and intuitive user interface.
Thesis (IT): https://docs.google.com/document/d/1KegpZaP6Rby7bHtz8KLOsaDHRtcslWYm-4ifaIQu2UA
Thesis (EN): coming soon...
This project was born from a real need. Then I realized that would be a nice idea to pick it up as a thesis for my Bachelor Degree.
This project lives in an Industry 4.0 context. Through analysis of data, coming from multiple sources, such as industrial machinery, this software can bring a real help and support for industry production. For instance, component failure prediction during manufacturing is one of the main goal of this project. As a result, it represents an added value to companies' business.
As of 16/03/2020, only Random Forest with Weka's default hyperparameters is used as a technique for ML models. Future improvements involves more techniques to be available for use.
This project is developed using web-based technologies. The software is usable through ordinary web browsers, even mobile devices. The back-end is written in Java, using Spring Boot 2 framework and H2 database engine. Tomcat Embedded is used during development to keep this software as lightweight as possible, but it's possible to export the project as a package to be deployed on any other application container.
It uses the version 3.6.14 of Weka. Don't use 3.8.0 and above because it leads to deadlock during static initialization (it calls Class.forName), which occurs in the com.github.fommil.netlib library, which comes as a dependency.