MATH-156-A5

Neural network implementation from scratch using SK learn In this project I implemented a Neural network from scratch with all the specifications provided in the brief As you can see loss decreases significantly through training, with the average cross validation loss set at 0.12482069692761051 Through varying the number of nodes in the hidden layer, learning rate, batch size and epochs. I have found a combination that yields a 95% f1 score which is high enough to deem it an appropriate model to apply to the provided dataset