Trying to guess is a project is doomed to fail, with the power of AI
The goal of this project is to train a MLP (Multi Layer Perceptron), using PyTorch, and try to predict, given a set of dependencies in a package.json file, if the project is error prone, by comparing it to other similar projects.
The (debatable) hypothesis here is that certain dependencies or set of dependencies combined together are more enclined to create bugs in a codebase than others.
At the moment, the weakness of the project is that the dataset is very thin. So this is an improvement opportunity.