Graduate Qualifying Project for the Data Science Course at BMSTU in Spring 2023 (11784DS)

In this grad project, I study the RUL dataset created by Ignacio Viñuales.

Results

This is the covariance matrix before and after preprocessing:

This shows how classic regressors sometimes perform better than a neural network:

This is a diagram of my dense neural network trained on the RUL dataset:

This is a Windows desktop client for the neural network. I used Microsoft's latest WinUI 3 framework and it is really cool. I also used the cpr - C++ Requests library and C++20 std::format function which are both very helpful for those with Python background.

My notebook for this dataset on Kaggle.