Project descriptions:
Project highlighting the strengths of the XGBoost algorithm. Full analysis of the alogrithm applied to the MNIST dataset, as well as the so-called
Perceptron module made from scratch. Backwards propogation implementation supports various learning rates, activation functions, cost functions, biases to reduce overfit. Perceptron module applied to bank data of credit card defaults. Model trained to predict credit card defaulting likelihood.
Regressional and statistical methods applied to fit and extend geological terrain data.
Smaller scale projects exploring various statistical phenomena